Acceptable Use Policy.
The rules on what may and may not be done with the HIPS Services, binding every Buyer, Member, Provider and Program Partner, what HIPS checks and when, and how HIPS enforces the rules and hears reports of abuse.
- Version
- 1.1
- Last updated
- In effect from
- 2026-11-16
- Length
- 15 sections17 defined terms
- Questions
- abuse@hips.comSection 15
This Acceptable Use Policy (the Policy) sets out what may and may not be done with the Services of HIPS Payment Group Limited (HIPS, we, us or our), which Hyperscale Inference Processing Systems OÜ operates for HIPS (Common Clauses Section 3.1), by Buyers, by the Members of their Workspaces, by Providers and by Program Partners (you or your); what HIPS checks, and when; the Measures HIPS takes when this Policy is breached and the reasons it gives for them; and how anyone can report abuse or illegal content. It is part of the Terms of Service, of the Provider Agreement and of the Affiliate Program Agreement (Section 1). Capitalized terms are defined in Section 2 and in Common Clauses Section 2.1.
Important: what HIPS may do when this Policy is breached
If HIPS has reasonable grounds to believe or suspect that this Policy has been breached, it may act at once and without prior warning: refuse or stop Requests and imports, remove a Customer Model, limit or pause a Workspace, stop assigning work to a Device, hold Earnings, Commissions and payouts as the Provider Agreement and the Affiliate Program Agreement allow, suspend or close Accounts, Workspaces, provider accounts and participations in the Affiliate Program, and report to the competent authorities (Section 12). HIPS tells you of each Measure and the reason for it, and you may ask a person at HIPS to review it (Section 13). Your liability for a breach of this Policy is outside the limits of liability of the Common Clauses (Common Clauses Section 5.9(e)). If you are a Provider, Section 9 forbids tampering with the Provider App, misreporting a Device, running more than one Device on one GPU, trying to recognize HIPS's Verification Work and inspecting what your Device processes; work done in breach of it earns nothing, as the Provider Agreement provides.
The commitments in brief
- Lawful use only. No child sexual abuse material, no intimate imagery of a real person without consent, no sexually explicit content designed for sexual gratification, no practice the EU AI Act prohibits, no help with weapons of mass destruction, no malware, fraud or other illegal content or use (Sections 5 and 6).
- Only models you have the right to run. A Customer Model is weight files imported from a model hub under a license that allows paid hosted inference, with a true attestation and a Hub Token you legitimately hold, and every model's license and acceptable-use policy binds you and your End Users (Section 7).
- One person, one identity, one Free Allowance. No second Account to gain a benefit or to evade a Measure, and no borrowed identities, self-referrals or self-dealing (Sections 4 and 8).
- Batches and prompt caches as designed. No splitting of Batches to evade limits or the Free Allowance, no Batch Files used as storage, and no probing of another Workspace's prompt cache (Sections 6.9 and 6.10).
- Honest referrals. In the Affiliate Program: no self-referral, no spam, a clear disclosure that you earn a Commission, only what HIPS publishes, no earnings claims, no bidding on the HIPS name, no shared Commissions or coupon sites, and no mention of any crypto-asset (Section 8.7).
- Honest GPU supply. Providers run the signed Provider App unmodified, report their Devices truthfully, enroll each machine once, as one Device, with every GPU it has reported, and never one GPU as two, never try to recognize Verification Work, never turn on a cache HIPS has turned off and never inspect what their Device processes (Section 9).
- No security testing without permission. Report vulnerabilities to HIPS instead; research that follows HIPS's Vulnerability Disclosure Policy is authorized. Under hips.com, the Services are only on ai.hips.com and its regional addresses, console.hips.com, providers.hips.com and the hosts under it, and weights.ai.hips.com, and the HIPS website is at hips.com, to which www.hips.com redirects (Section 10).
- HIPS never trains on, retains or logs your content. HIPS detects abuse from metadata, rate signals, measurements and reports, and keeps no content beyond what you ask it to hold to deliver what you asked for (Common Clauses Section 3.14). A person at HIPS reads content that HIPS holds for you only where the law requires it, with a second administrator's approval and a record, and HIPS keeps no content of a flagged Request unless the law requires it. A catalog model that can generate realistic images, video or audio of people runs only behind automated safeguards against child sexual abuse material and intimate imagery made without consent (Section 11).
- Objective grounds, reasons and review. HIPS acts on objective grounds and in proportion, tells you the reasons for a Measure, and lets you ask a person to review it (Sections 3, 12 and 13).
- Anyone can report. Illegal content and abuse can be reported through HIPS's notice form, and HIPS decides on every Notice (Section 14).
01This Policy and its place in the Agreement
What this Policy is. This Policy sets out the rules on the use of the Services. It applies to every use of the Services, through the Console, the API, the Provider App, a tool you authorize through the Console, or any other means.
Part of the Terms of Service, the Provider Agreement and the Affiliate Program Agreement. This Policy is incorporated by reference into the Terms of Service, into the Provider Agreement and into the Affiliate Program Agreement, and forms part of each of them as if it were written out in it. When you accept any of them, you accept this Policy as part of it, and HIPS records the version of this Policy with your acceptance. A breach of this Policy is a breach of the Agreement of which it forms part. You do not accept this Policy on its own.
The Common Clauses. This Policy is an Incorporating Document of the Common Clauses and incorporates version 1.1 of them by reference, word for word, and they form part of it as if they were written out in it (Common Clauses Section 1.1). Among them, their definitions (Common Clauses Section 2), their rules on sanctions and export control (Common Clauses Section 7), changes (Common Clauses Section 9), notices (Common Clauses Section 10), disputes (Common Clauses Sections 12 to 15), severability (Common Clauses Section 16) and survival (Common Clauses Section 19) apply to this Policy.
Who this Policy binds. This Policy binds every person who accepts the Terms of Service, the Provider Agreement or the Affiliate Program Agreement, and every Member in their use of a Workspace, as the Terms of Service provide. Its Sections apply as follows:
- (a)Sections 1 to 4 and 10 to 15 apply to everyone;
- (b)Sections 5, 6 and 7 apply to the use of the Services as a Buyer or a Member;
- (c)Section 8 applies to everyone who receives the Free Allowance, Credits, a promotion or a Referral Discount, and to everyone who takes part in the Affiliate Program or uses a Referral Link or a Referral Code, from their first use, whether or not they have accepted the Affiliate Program Agreement; and
- (d)Section 9 applies to Providers.
A person who is a Buyer, a Provider or a Program Partner in more than one of these capacities is bound in each of them.
Responsibility for others: Business Users and US Users. If you are a Business User or a US User, you are responsible for every use of the Services through your Account, your Workspaces, your API Keys, your Tool Tokens and your Devices, by your Members, your End Users, your employees and contractors and anyone else, as for your own use, unless the use results from HIPS's breach of the Agreement.
Responsibility for others: Consumers. If you are a Consumer, you are responsible for a use of the Services through your Account, your API Keys, your Tool Tokens or your Devices by another person if you allowed it, or if it was made possible because you did not keep your credentials secure as Section 4.5 requires.
End Users. If you make the Services, or applications that use them, available to End Users, you shall:
- (a)bind your End Users to terms that prohibit at least what this Policy prohibits for the use they make, and to the use restrictions of the license of each model they use, as Section 7.5 and Terms of Service Section 23.2 describe, and tell them that those restrictions apply;
- (b)take reasonable steps to prevent and to stop their breaches;
- (c)act without undue delay on HIPS's request concerning an End User's breach, including by suspending that End User's access; and
- (d)never give an End User an API Key or a Tool Token.
Relationship with the law and the other documents. This Policy adds to Applicable Law and to the rest of the Agreement. Complying with this Policy does not make lawful what Applicable Law forbids, and a use that Applicable Law forbids is prohibited whether or not this Policy names it. A rule of this Policy that is stricter than a rule of the Terms of Service, the Provider Agreement or the Affiliate Program Agreement on the same conduct does not conflict with it, and both apply. Any other conflict between documents is resolved by Common Clauses Section 17.3.
Changes. HIPS changes this Policy only under Common Clauses Section 9, including for the reasons Common Clauses Section 9.1(a) and (d) give: a change in Applicable Law, and a security, fraud-prevention or abuse-prevention need. A change that addresses an unforeseen and imminent risk of fraud, malware, spam, a data breach or another security risk, or that Applicable Law requires sooner, may take effect sooner, as Common Clauses Section 9.3 provides.
Survival. In addition to the provisions that Common Clauses Section 19 preserves, Sections 9.6(c), 9.7, 12.8 to 12.12 and 13 of this Policy survive the end of the Agreement, the closure of an Account, a Workspace or a provider account, and the end of a participation in the Affiliate Program.
02Definitions and interpretation
Defined terms. Capitalized terms that this Policy does not define have the meanings given in Common Clauses Section 2.1, among them Batch, Batch File, Batch Line and Completion Window. In addition, in this Policy:
- AI Act
- means Regulation (EU) 2024/1689 on artificial intelligence, as amended, including by Regulation (EU) 2026/1744.
- API Key, Cached Input Token, Free Allowance, Hub Token, Prefix Ledger and Promotional Credits
- have the meanings given in Section 2.2 of the Terms of Service, Prompt Data the meaning given in Common Clauses Section 2.1, and a prompt cache is what Section 9.9 of the Terms of Service calls one.
- Commission, Referral, Referral Code, Referral Link and Referred Person
- have the meanings given in Section 2.1 of the Affiliate Program Agreement, and Referral Discount the meaning given in Section 2.2 of the Terms of Service; Affiliate Program, Affiliate Program Agreement and Program Partner have the meanings given in Common Clauses Section 2.1.
- Digital Services Act
- means Regulation (EU) 2022/2065 on a Single Market for Digital Services.
- End User
- means a person who uses an application, product or service that a Buyer builds or offers with the Services, other than as a Member of the Buyer's Workspace.
- Illegal Content
- means any information that, in itself or in relation to an activity, does not comply with Applicable Law, including the law of the European Union or of any of its Member States.
- Import
- means HIPS's process of fetching a Customer Model from a model hub at a pinned revision, checking its files, re-saving and copying the files it accepts into HIPS storage, and signing them.
- Invitation Code
- means the code HIPS sends by email to a person whose early-access registration it approves, and which that person needs to create an Account.
- Measure
- means any action HIPS takes under Section 12 that restricts the use of the Services, an Account, a Workspace, an API Key, a Tool Token, a Customer Model, a Device, a provider account, Credits, Earnings, a Commission, a participation in the Affiliate Program or a payout.
- Notice
- means a report of Illegal Content or of a breach of this Policy made under Section 14, and Notifier means the person who makes it.
- Security Research
- means testing the security of the Services or of any system connected to them, including probing, scanning, fuzzing, and attempting to bypass authentication, authorization, a sandbox, limits or metering, or to reach data or systems.
- Statement of Reasons
- means the statement HIPS gives under Section 13.
- Tool Token
- means an access token or a refresh token that HIPS issues to a tool a person authorizes through the Console, such as the Provider App.
- Verification Work
- means the work HIPS assigns to a Device to measure it and the work it performs, including jobs whose correct result HIPS already knows, probes of a Device's memory and capacity, completed jobs sampled and run again for comparison, in memory only, on a GPU that HIPS itself operates (Section 11.6), and decoy models. Verification Work is not marked, and it reaches a Device by the same path as other work.
- Verified Person
- means the natural person whom HIPS's identity verification establishes as the person behind an Account. Every Account that identity verification resolves to the same natural person, and every Workspace and provider account that the person holds through such an Account as Buyer or as Provider, belongs to that Verified Person. A Workspace held by another Buyer, or a provider account held by another Provider, does not belong to a Verified Person because they are a Member of it or acts for it; a Measure concerning that person affects that Workspace or provider account only by suspending their membership or authority to act for it, unless Section 12.1 gives HIPS a ground concerning that Workspace or provider account itself.
- Vulnerability Disclosure Policy
- means the policy on reporting vulnerabilities and on good-faith security research that HIPS publishes at https://hips.com/security (Section 10.5).
- Workload
- means any Request, job, deployment, Import or other processing you ask the Services to perform, and, for a Provider, any work HIPS assigns to a Device.
Interpretation. Common Clauses Section 2.2 applies to this Policy. In addition:
- (a)a list of examples illustrates a rule; it does not narrow the rule, and it does not narrow what Applicable Law prohibits;
- (b)a prohibition on doing something includes knowingly attempting it, and helping, encouraging or allowing anyone else to do it;
- (c)a prohibition on content covers content in an Input, an Output, a Customer Model, an uploaded file, a job result and anything else you submit to or obtain from the Services;
- (d)"reasonable grounds to believe or suspect" is an objective standard: it means grounds that would lead a reasonable operator of the Services, on the information HIPS has, to believe or to suspect the matter concerned;
- (e)a reference to a Section is to a Section of this Policy, unless the Common Clauses or another document is named; and
- (f)"Affiliate" on its own has the meaning Common Clauses Section 2.1 gives it, a company in HIPS's group. The Affiliate Program is the name of a program, and a Program Partner is not an Affiliate of HIPS.
03How HIPS applies this Policy
Objective grounds. HIPS takes a Measure only where it has reasonable grounds to believe or suspect that a breach of this Policy, of the Agreement or of Applicable Law has occurred, is occurring or is about to occur, or where Applicable Law or an order of a competent authority requires it.
Diligent, objective and proportionate. HIPS applies and enforces this Policy in a diligent, objective and proportionate manner, with due regard to the rights and legitimate interests of everyone involved, including the fundamental rights of the people concerned, such as freedom of expression and the protection of personal data, as Article 14(4) of the Digital Services Act requires.
What HIPS weighs. In choosing a Measure and its scope, HIPS considers:
- (a)the nature and seriousness of the breach and the harm it causes or risks to people, to Buyers, to Providers, to HIPS and to the Services;
- (b)whether the breach was intended, repeated or concealed;
- (c)the strength of the evidence and of the signals;
- (d)the risk that the breach continues, or spreads to other Accounts, Workspaces or Devices;
- (e)your cooperation, and any remedy you have made; and
- (f)what Applicable Law requires.
No duty to monitor. HIPS has no general obligation to monitor the content it processes or stores, or to search actively for facts or circumstances indicating illegal activity, and it does not undertake to do so. HIPS enforces this Policy to protect the Services, the people who use them and the public. Neither its enforcement nor a decision not to act in a particular case creates a duty owed to you, except where Applicable Law imposes one.
HIPS's records. The facts on which HIPS acts may be established from HIPS's records, which have the effect Common Clauses Section 17.4 gives them.
Confidential controls. HIPS does not publish the rates, thresholds, weights, test inputs, references or methods of its fraud, abuse and verification controls, or which jobs are Verification Work, because disclosing them would let them be defeated. A Statement of Reasons describes the ground and the facts relied on without disclosing them (Section 13.4). HIPS nevertheless explains to a person, on request, which of their personal data it used and how they led to a decision about them, and gives the information it does not publish to a supervisory authority or a court that asks for it, as Case C-203/22 requires.
04General rules
Lawful use. You shall use the Services only in compliance with Applicable Law and the Agreement, and not in a way that infringes the rights of others.
True information. You shall give HIPS true, accurate and complete information, and keep it current, including in identity verification, in your declaration of status and Country, and in your business, tax and payout information. You shall not:
- (a)present to identity verification another person's identity document, face or details, or images, video or documents that are synthetic, altered or replayed;
- (b)misrepresent your Country or your location, including by a virtual private network, a proxy or any other means (Common Clauses Section 7.6); or
- (c)declare yourself a Business User when you know that you act as a Consumer, or a Consumer when you know that you act as a Business User, or support your declaration of status with untrue facts, such as an organization, a trade name, or a registration or VAT number that is not yours.
A finding by a court or tribunal that you act in the other capacity does not, on its own, make your declaration of status a breach of (c). A declaration of status whose wording HIPS required as a condition of contracting, without an option to declare otherwise, is not a breach of this Section 4.2.
One person, one identity. Each natural person may hold one verified identity. One Account may own or belong to several Workspaces, as the Terms of Service allow. You shall not:
- (a)open or use more than one Account to obtain a further Free Allowance, a promotion or any other benefit, or to evade a limit, a Measure, a pause or a closure;
- (b)let another person use your Account, your sign-in methods or your identity;
- (c)use another person's Account, or an Account opened with a false, borrowed or purchased identity; or
- (d)open an Account, a Workspace or a provider account to escape a Measure, a limit, a pause or a closure.
Invitations. Invitation Codes and invitations to a Workspace are personal to the person to whom HIPS or the Workspace sends them. You shall not sell, transfer, publish or share them, or use one sent to someone else.
Credentials. You shall keep your sign-in methods, API Keys, Tool Tokens and Hub Tokens secret and secure, and you shall not:
- (a)publish an API Key, or embed it in software or a web page that runs in a browser or that you distribute to your customers or other users (Terms of Service Section 5.2);
- (b)share a credential with anyone who is not entitled to use it under the Agreement; or
- (c)use a credential that was not issued to you, to your Workspace or to a tool you authorized.
You shall revoke a credential, and tell HIPS, as soon as you suspect that it has been exposed or misused. HIPS may revoke an API Key, a Tool Token or a Device certificate that it has reasonable grounds to believe has been exposed or misused, and tells you when it does.
Sharing and resale of access. You may build and offer your own applications, products and services that use the Services, and use their Outputs, as the Agreement allows. Unless HIPS agrees in writing, you shall not:
- (a)sell, rent, lend or otherwise give access to your Account, a Workspace, an API Key or a Tool Token to anyone who is not a Member of the Workspace, except an API Key that you give, as Terms of Service Section 5.2 allows, to an employee, worker, contractor or service provider of yours who uses it only on your behalf and is bound to keep it confidential; or
- (b)operate a service whose main purpose is to resell access to the Services, or to the models HIPS serves, as such, to third parties.
Interference and limits. You shall not interfere with or disrupt the Services, the Devices of Providers, the capacity of Approved Providers or the use of the Services by others. In particular, you shall not:
- (a)send traffic designed to exhaust capacity, or to force repeated model loads, cancellations or retries without a genuine use;
- (b)evade, or try to evade, the rate limits, concurrency limits, fairness rules, budgets or metering of the Services; or
- (c)use several Accounts, Workspaces or API Keys to exceed a limit that applies to one.
Identifying other users. HIPS does not tell Buyers and Providers each other's identity (Common Clauses Section 3.8). You shall not use the Services to identify, locate, fingerprint or contact any Provider, Device or Buyer, or to find out which Device or Workspace processed or sent a Request.
Sanctions and export control. Common Clauses Section 7 applies to every use of the Services. In particular, you shall not give the Government of Russia, a legal person, entity or body established in Russia, or the Republic of Belarus, its Government or one of its public bodies, corporations or agencies, access to the Services or to any model, inference or computing capacity you obtain through them (Common Clauses Section 7.3), and you shall not use the Services or an Output for an end use that Export Control Laws prohibit (Common Clauses Section 7.5).
Automated access. You shall access the Services only through the interfaces HIPS provides, and use them as HIPS documents them. You shall not collect data from the Console or the HIPS website by scraping or other automated means, except through the API as documented.
Cooperation. You shall answer HIPS's reasonable requests for information about a suspected breach of this Policy that concerns your Account, your Workspaces, your End Users or your Devices, within the reasonable time HIPS sets, and keep the information concerned while HIPS investigates.
05Prohibited content
Scope. You shall not submit content that this Section describes to the Services, use the Services to generate it, or use, keep, publish or distribute an Output that is such content. If a model generates such content without your asking for it, you shall not use, keep or distribute it, except to report it to HIPS or to a competent authority.
Important: no child sexual abuse material, on any Plan, with any model
Child sexual abuse and the sexualization of minors. This Section applies to everyone, on every Plan, with every model and every Customer Model, without exception. You shall not use the Services for:
- (a)child sexual abuse material, including any material or performance within points (c) and (e) of Article 2 of Directive 2011/93/EU and any other material Applicable Law treats as child sexual abuse material, whether it is real, realistic, generated, manipulated, drawn or fictional;
- (b)sexual content that depicts or describes a person under eighteen (18) years of age, or a person presented as under eighteen (18);
- (c)content that grooms, solicits, sexually exploits or extorts a minor, or that facilitates the trafficking or exploitation of a minor; or
- (d)a Customer Model, or any other file, designed, trained or fine-tuned to generate content described in (a) or (b).
HIPS treats a breach of this Section as a serious breach, may close every Account of the Verified Person concerned with immediate effect, and reports the content, and the information needed to identify who submitted it, to the competent authorities and organizations as Applicable Law requires or permits, including: (a) under Article 18 of the Digital Services Act, the law enforcement or judicial authorities of the Member State concerned, or, where HIPS cannot identify that Member State with reasonable certainty, the law enforcement authorities of Ireland, where HIPS is established, or Europol, or both; (b) any authority whose order under Article 9 or 10 of the Digital Services Act requires it; and (c) where 18 U.S.C. 2258A applies to HIPS, the CyberTipline of the National Center for Missing & Exploited Children. HIPS preserves the content and the records concerned for as long as the law under which it reported requires, including one (1) year after a report to the CyberTipline (18 U.S.C. 2258A(h)), and uses them for no other purpose.
Intimate imagery without consent. You shall not use the Services to generate, manipulate, distribute or threaten to distribute realistic images, video, audio or similar material of an identifiable natural person's intimate parts, or of an identifiable natural person engaged in sexually explicit activity, without that person's freely given, specific, informed, unambiguous and explicit consent to that generation or manipulation. From 2 December 2026, point (ba) of Article 5(1) of the AI Act prohibits AI systems and uses of this kind, as that point and Article 5(1a) describe them; this Policy prohibits every such use of the Services from the day you accept it.
Sexually explicit content. You shall not use the Services, on any Plan and with any model or Customer Model, to generate, request, store or distribute sexually explicit content, being content that depicts or describes sexual acts, or nudity in a sexual context, in explicit detail and that is designed for the purpose of sexual gratification, including pornography, erotic role-play and sexual chatbots, whether the people it depicts or describes are real or fictional. This Section does not prohibit content that serves a medical, health, educational, scientific, documentary, artistic or news purpose and is not designed for sexual gratification, such as sex education or a clinical description. You shall not import, deploy or use a Customer Model designed, trained or fine-tuned mainly to generate sexually explicit content. Sections 5.2 and 5.3 apply in every case.
Violence, terrorism and threats. You shall not use the Services for content that:
- (a)incites, threatens or calls for violence against a person or a group;
- (b)incites, solicits, instructs or glorifies the commission of a terrorist offense, or recruits for, finances or supports a terrorist group, within the meaning of Applicable Law;
- (c)gives operational help in planning or carrying out a mass-casualty attack; or
- (d)depicts real violence, and is made or distributed to glorify it or to intimidate.
This Section does not prohibit content that reports on, documents, analyzes, counters or researches these matters for journalistic, educational, artistic, academic, safety or law-enforcement purposes, as long as the content does not itself incite, threaten or give operational help.
Hate, harassment and harm to individuals. You shall not use the Services for content that:
- (a)incites hatred, discrimination or violence against a person or a group because of race, color, ethnic or national origin, religion or belief, disability, sex, sexual orientation, gender identity, age or any other characteristic that Applicable Law protects;
- (b)harasses, bullies, intimidates or stalks an identifiable person;
- (c)publishes personal data of a person in order to expose them to harassment or harm;
- (d)is defamatory of a person or an organization; or
- (e)encourages a specific person to commit suicide or to harm themselves, or gives methods of self-harm to a person at risk, except in the context of prevention, treatment or research.
Weapons and dangerous materials. You shall not use the Services for content that:
- (a)gives meaningful help in developing, producing, acquiring, stockpiling, enhancing or using chemical, biological, radiological or nuclear weapons, or any other weapon capable of causing mass casualties, or their means of delivery;
- (b)gives instructions for making explosive devices, or improvised weapons for use against people or property; or
- (c)facilitates the illegal manufacture, conversion, acquisition or trafficking of firearms, ammunition or their essential parts.
Malicious code and attacks on systems. You shall not use the Services:
- (a)to create, improve or distribute malware, ransomware, spyware, botnets, exploits or other code designed to damage, disable, gain unauthorized access to or take control of a system, a device or data;
- (b)to plan or carry out unauthorized access to, intrusion into, or a denial-of-service or other attack against any system, network or person; or
- (c)to create phishing material, or to carry out credential stuffing or the theft of credentials.
This Section does not prohibit authorized security testing of systems you are entitled to test, defensive research or education that gives no working attack capability against systems you are not entitled to test. Section 10 governs any testing of the Services themselves.
Fraud, deception and spam. You shall not use the Services for:
- (a)fraud, scams, phishing or impersonation for gain, or the forgery of documents, including identity documents, receipts, invoices, reviews, certificates and records presented as genuine;
- (b)unsolicited commercial communications sent in breach of Applicable Law, or bulk messages that their recipients have not requested;
- (c)fake reviews, fake accounts or coordinated inauthentic behavior designed to mislead people about who is speaking or how many people hold a view;
- (d)deceiving people about the time, place, means or eligibility of voting, or otherwise interfering with an election or a referendum in breach of Applicable Law; or
- (e)pyramid schemes, market manipulation, insider dealing, money laundering, terrorist financing, or the evasion of sanctions or of taxes.
Impersonation and deep fakes. You shall not use the Services:
- (a)to impersonate a real person, an organization or an authority, HIPS included, in a way designed to mislead; or
- (b)to publish or distribute image, audio or video content generated or manipulated by artificial intelligence that resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful, without disclosing that it was artificially generated or manipulated, where Article 50(4) of the AI Act or other Applicable Law requires that disclosure.
Privacy and surveillance. You shall not use the Services:
- (a)to process personal data in breach of data protection law, including without a lawful basis or without the information the law requires you to give;
- (b)to process special categories of personal data, or personal data relating to criminal convictions and offenses, without the condition Applicable Law requires;
- (c)to collect, infer or compile information about individuals for unlawful surveillance, tracking, profiling or monitoring, including covert monitoring of a person's communications, location or devices; or
- (d)to identify an individual from images, video or audio of their face, body or voice, except where Applicable Law allows it and you meet its conditions, and never for a practice Section 6.1 describes.
Intellectual property and confidential information. You shall not use the Services for content that infringes or misappropriates a copyright, trademark, patent, trade secret, database right, or right of publicity or personality, or that discloses confidential information you have no right to disclose. You shall not use the Services to circumvent technological protection measures, or to remove rights-management information, in breach of Applicable Law.
Other illegal goods, services and content. You shall not use the Services for:
- (a)the sale or promotion of illegal drugs, weapons, counterfeit or stolen goods, human organs, or any other goods or services whose sale Applicable Law prohibits;
- (b)human trafficking, forced labor or sexual exploitation;
- (c)gambling that Applicable Law prohibits where it is offered; or
- (d)any other Illegal Content.
06Prohibited uses and workloads
Practices the AI Act prohibits. You shall not use the Services, a Customer Model or an Output for any practice that Article 5 of the AI Act prohibits, whether or not the AI Act applies to you, including:
- (a)techniques that deploy subliminal, purposefully manipulative or deceptive methods to distort a person's behavior materially, impairing their ability to make an informed decision, in a way that causes or is reasonably likely to cause significant harm;
- (b)exploiting the vulnerabilities of a person or a group due to age, disability or a specific social or economic situation, to distort their behavior materially in a way that causes or is reasonably likely to cause significant harm;
- (c)social scoring: evaluating or classifying people over time on the basis of their social behavior or personal characteristics, where the resulting score leads to detrimental or unfavorable treatment in unrelated contexts, or treatment that is unjustified or disproportionate;
- (d)assessing or predicting the risk that a person will commit a criminal offense based solely on profiling or on an assessment of personality traits and characteristics, except to support a human assessment based on objective and verifiable facts directly linked to a criminal activity, as Article 5 permits;
- (e)creating or expanding facial recognition databases through the untargeted scraping of facial images from the internet or from closed-circuit television footage;
- (f)inferring the emotions of a person in the workplace or in an education institution, except for medical or safety reasons as Article 5 permits;
- (g)biometric categorization that categorizes people individually on the basis of their biometric data to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation, except the labeling or filtering of lawfully acquired biometric datasets where Article 5 permits it;
- (h)real-time remote biometric identification of people in publicly accessible spaces for the purposes of law enforcement, which this Policy prohibits through the Services in every case;
- (i)generating or manipulating intimate material of an identifiable person without consent, as point (ba) of Article 5(1) describes (Section 5.3); and
- (j)generating or manipulating material or a performance within points (c) and (e) of Article 2 of Directive 2011/93/EU, as point (bb) of Article 5(1) describes (Section 5.2).
This list summarizes Article 5. Where it differs from Article 5, Article 5 decides what is prohibited, except that this Policy prohibits (h) without exception, and prohibits (i) and (j) before the date from which the AI Act applies them.
Your obligations under the AI Act. If you build, place on the market, put into service or use an AI system with the Services, you are responsible for the obligations that the AI Act and other Applicable Law place on you as its provider, deployer, importer or distributor, including:
- (a)designing a system that interacts directly with people so that they are informed that they are interacting with an AI system, where Article 50(1) requires it;
- (b)disclosing deep fakes, and text published to inform the public on matters of public interest, as artificially generated or manipulated, where Article 50(4) requires it;
- (c)the obligations that apply to a high-risk AI system; and
- (d)the measures on AI literacy that Article 4 requires of you.
HIPS places on the market, under its own name, the AI systems through which it serves catalog models through the API and the Console, and is their provider within the meaning of Article 3(3) of the AI Act; the publisher of each model remains the provider of that model. HIPS marks the Outputs of those systems that are synthetic audio, images, video or text in a machine-readable format, and makes them detectable as artificially generated or manipulated, as Article 50(2) of the AI Act requires, as far as this is technically feasible and except where an exception of that Article applies, such as for source code and other structured output, very short output, and output that does not substantially alter the Input you provided. The page of each catalog model and HIPS's documentation describe how its Outputs are marked and how a mark can be detected. For a Customer Model, you choose the model, import it and decide how it is used, and you are the provider or the deployer of any AI system you build with it; HIPS runs it on your instructions and does not mark its Outputs unless the model's page says otherwise. The Agreement does not transfer HIPS's own statutory duties to you, and you do not transfer yours to HIPS.
Markings and provenance. You shall not remove, alter, obscure or defeat a watermark, a machine-readable marking, provenance metadata or any other indication that content was artificially generated or manipulated, that HIPS or a model applies to an Output. You shall not present an Output as made by a human where Applicable Law requires its artificial origin to be disclosed, or in order to deceive.
Decisions about people. You shall not use an Output as the sole basis for a decision that produces legal effects concerning a natural person or similarly significantly affects them, such as a decision on credit, employment, housing, insurance, education, health care, immigration, or access to essential services or public benefits, unless Applicable Law permits that decision and you meet the conditions it sets, including human review where it requires one. Common Clauses Section 4.7 applies.
Safety-critical and military uses. Unless HIPS agrees in a signed document, you shall not use the Services or an Output:
- (a)to operate or control, in real time and without human oversight, a system whose failure could lead to death, personal injury or severe physical or environmental damage, such as a medical device, life support, a vehicle, aircraft or air traffic control, a nuclear facility or critical infrastructure; or
- (b)to design, develop, produce or operate weapons, or for military targeting.
Regulated services. You shall not use the Services to provide a regulated service, such as a legal, medical, financial, investment, insurance or tax service, without the authorization, license or qualification that Applicable Law requires of you. You shall not present an Output as advice from a qualified professional or from HIPS (Common Clauses Section 4.6).
Model extraction and distillation. You shall not:
- (a)use the Services to extract, reconstruct, copy or distill the weights, architecture or behavior of a model in breach of its license, its terms of use or Applicable Law;
- (b)try to obtain the weights of a Customer Model of another Workspace, or of any model that the Services do not make available to you; or
- (c)try to obtain information the Services do not show you, such as other users' prices, Providers' minimum prices, HIPS's floors and rates, or the identity of a Device or a Workspace.
You may distill your own Customer Model through your own Workspace, within its license. HIPS adds no restriction of its own on using the Outputs of a catalog model to develop other models beyond items (a) and (c), the license of the model concerned (Terms of Service Sections 16.4 and 23.2) and Applicable Law.
No training data. HIPS never trains on Prompt Data, for anyone and on any Plan (Common Clauses Section 3.14(a)): no Input or Output of a Workspace, no Batch Line and nothing HIPS holds at a Buyer's request is used to train, fine-tune, distill or evaluate any model, and no Request is screened, kept or curated for that purpose. Section 6.7 governs the use of Outputs to develop other models, and Section 6.10 the poisoning of prompt caches.
Batches. Each Batch Line is a Request, and this Policy applies to it, and to every Batch File, as it applies to any other Request and Input; HIPS may refuse or cancel a whole Batch whose lines breach this Policy. In addition, you shall not:
- (a)split a workload across Batches, Workspaces, Accounts or API Keys, or create, cancel and create Batches again, in order to exceed or evade a limit that applies to Batches, such as the number of lines or the size of an input file, the number of Batches a Workspace may create in an hour or have open at once, or the input tokens it may have waiting for a model (Terms of Service Section 6.16), or any other limit of the Services;
- (b)use Batches to obtain Outputs beyond what the Free Allowance or a promotion of a Workspace covers, including by spreading Batches across the Workspaces or Accounts of the same Verified Person or of persons acting together, or to use the Free Allowance, Promotional Credits or held Credits of one Workspace for the benefit of another person or Workspace;
- (c)create Batches that you do not intend to use, in order to occupy capacity, to keep Credits held, to delay or block the Requests of others, or to probe prices, capacity or the placement of work;
- (d)use a Batch File, or the file interface of the API, to store, host, share or distribute content for any purpose other than submitting a Batch and retrieving its results, or upload a file for a purpose the API does not give it;
- (e)design a Batch, a Batch Line or a Completion Window so that Outputs are delivered without being charged, such as by relying on lines that run but are reported as expired or canceled, or by exploiting a fault in the metering, the held maximum or the expiry of Batches; or
- (f)include in a Batch a line that HIPS would refuse as a single Request on a ground other than the available Credits (a Batch whose maximum charge the available Credits cannot carry ends as failed under Terms of Service Section 6.21, which is no breach of this Policy), or that refers to a file, a Customer Model, a deployment or a prompt cache of another Workspace.
A fault in the metering, the limits, the held maximum or the expiry of Batches that you discover is a vulnerability that you shall report under Section 10.5 and not use.
Prompt caches. HIPS keeps prompt caches apart for each Workspace (Terms of Service Section 9.9). You shall not:
- (a)try to learn whether, or what, another Workspace or another person has sent, or what a prompt cache of another Workspace holds, including by measuring the time, the cost or the number of Cached Input Tokens of your Requests, by guessing or reusing the cache keys of others, or by any other side channel;
- (b)try to read, use, share, alter, poison or evict a prompt cache of another Workspace, or to make a prompt cache made from your Requests serve another Workspace;
- (c)set, or try to set, the value with which HIPS keeps the prompt caches of Workspaces apart, or otherwise try to defeat that separation;
- (d)send Requests designed to fill, flush or exhaust prompt caches, or to force the placement of your Requests on a particular worker or Device, without a genuine use;
- (e)manipulate the count of Cached Input Tokens, or try to be charged the cached-input rate for input that was not read from a prompt cache of your Workspace; or
- (f)use the behavior of prompt caches to identify or fingerprint a Device, a worker, a pool or a Provider (Section 4.8).
Testing the separation of prompt caches is Security Research and needs HIPS's authorization under Section 10.3. Measuring the latency and the Cached Input Tokens of your own Requests within your own Workspace is not (Section 10.2).
07Customer Models and model licenses
Weight files only. A Customer Model is imported from a model hub that HIPS supports and consists of weight files that HIPS's own engines run. No code you supply runs anywhere. HIPS refuses at Import, among other things:
- (a)code, including repository code that a model would need to run, a configuration that would load code, and custom operators;
- (b)pickle-based files, and any file whose bytes are a pickle, whatever its name;
- (c)external data that points outside the model's own files; and
- (d)a chat template, tokenizer file or configuration that HIPS's checks do not accept.
An Import that HIPS refuses for one of these reasons is not, on its own, a breach of this Policy. You shall not, however, import or attempt to import a file built to exploit, crash or exhaust the Import, a parser, an engine, a worker or a Device, or a model built to read, collect or send out data, files or secrets from any machine that runs it; and you shall not repeat, or vary in order to pass HIPS's checks, an Import that HIPS has refused for one of the reasons in (a) to (d).
Your right to run the model. At every Import you represent and warrant, and the Member who imports attests, that:
- (a)you have the right to have HIPS store the model and run it for you as a paid hosted service, and to use its Outputs as you intend, under its license and every other term that binds it;
- (b)where the hub gates the repository, the hub account whose Hub Token you use has accepted the gate's terms, and those terms allow that use;
- (c)where you rely on HIPS's exception for your own model, the repository belongs to the hub account your Hub Token signs in as, and every base model it derives from is under a license HIPS accepts; and
- (d)as far as you know, the information the hub shows about the model, including its license, is not false.
HIPS imports a model only under a license on the list HIPS accepts for paid hosted inference, as the import screen of the Console shows it, except your own model under (c). On the date of this Policy, the list holds, in the identifiers Hugging Face states: apache-2.0, mit, bsd-2-clause and bsd-3-clause; the Llama community licenses llama2, llama3, llama3.1, llama3.2 and llama3.3, for models that take text input only, for every Buyer; the Gemma terms, gemma; and, only as the next sentence allows, llama4, and llama3.2 for a multimodal model. A model under the llama4 license, or a multimodal model under the llama3.2 license (one whose model card shows image input), including a model derived from one, is never part of HIPS's catalog, and HIPS imports it only as a Customer Model of a Buyer that is established and domiciled outside the European Union and that represents and warrants at the import, as the Member who imports attests, that it holds Meta's license for that model in its own name and is neither a natural person domiciled in the European Union nor a company with its principal place of business in the European Union; HIPS runs such a model only in its regional processing locations outside the European Union, never in the `eu` region and never for a Buyer established or domiciled in the European Union (Terms of Service Sections 6.8 and 15.4). Otherwise, HIPS refuses a model whose license does not allow it to be run as this Section and Terms of Service Section 15.4 provide. For a gated repository, HIPS records the hub account your Hub Token signs in as, where the hub states it, and the time the Import read the gated files. The license the hub shows is the declaration of the repository's owner, which HIPS records but cannot verify; the attestation is your own. The Member who imports gives it in these words, in the version the import screen shows:
"I confirm, for this Workspace and with its authority, that: (1) we have the right to have HIPS store this model at this revision and run it for us as a paid hosted inference service, and to use its Outputs as we intend, under its license, its acceptable-use policy and the terms of any gate our hub account accepted; (2) we may use the Hub Token we give HIPS for this import, and the hub account it signs in as accepted any gate; (3) as far as we know, what the hub shows about this model, its license included, is true; (4) the model contains no personal data that we may not have processed by HIPS; (5) the model is not designed, trained or fine-tuned to generate child sexual abuse material, intimate imagery of a real person without consent, sexually explicit content, or anything else that the Acceptable Use Policy or Article 5 of the EU AI Act prohibits; (6) we will comply with the model's license in using it and its Outputs, including its attribution, notice and acceptable-use terms, and bind our End Users to its restrictions; and (7) this confirmation repeats the warranty of Terms of Service Section 15.4, which binds the Buyer."
For a model under the llama4 license, or a multimodal model under the llama3.2 license, the import screen adds to the attestation: "We are established and domiciled outside the European Union, we are neither a natural person domiciled in the European Union nor a company with its principal place of business in the European Union, and we hold Meta's license for this model in our own name."
Hub Tokens. You shall give HIPS only a Hub Token that you, or the organization you act for, legitimately hold and are entitled to use for the Import. You shall not use a Hub Token that was leaked, found, shared with you without the authority of the account holder, or taken from another person, and you shall not import a private or gated repository you are not authorized to read. HIPS stores read-only Hub Tokens only. HIPS cannot revoke a Hub Token at its hub, so you shall revoke it there when you no longer need it or suspect that it has been exposed.
Using Customer Models within their licenses. You shall comply with the license and the terms of each Customer Model in your use of the model and of its Outputs, including duties of attribution and notice, restrictions on fields of use, on users, on volumes and on commercial use, the model's own acceptable-use policy, and any duty to pass terms on to your End Users.
Catalog models. The page of each catalog model states the license and usage conditions that bind your use of the model and of its Outputs, and you shall comply with them. For a model under a Llama community license, Meta's Acceptable Use Policy for that version, and for a model under the Gemma Terms of Use, the Gemma Prohibited Use Policy, are part of those conditions. A use that a catalog model's license or acceptable-use policy prohibits is prohibited under this Policy too.
When your right ends. If your right to run a Customer Model ends or is credibly disputed, including because the license under which you imported it is terminated for you, a rights holder objects, the hub removes the repository, or the gate's terms are withdrawn for you, you shall stop the model's deployments, tell HIPS, and delete the Customer Model or the version concerned without undue delay.
HIPS's checks are not a clearance. HIPS's acceptance of an Import, the license it records and the checks it runs do not mean that a model is lawful, safe, licensed for your use or fit for any purpose, and HIPS does not verify what a hub states (Common Clauses Section 4.5(d)). You remain responsible for your rights to what you import (Common Clauses Section 6.1(a)).
Models that generate prohibited material. HIPS may refuse, stop or remove a Customer Model, and refuse Requests to it, where it has reasonable grounds to believe that the model is designed, trained or fine-tuned to generate content that Section 5.2, 5.3 or 5.4 describes, or that its design makes such generation a reasonably foreseeable and reproducible outcome that its deployment does not reliably prevent.
Rights holders. A rights holder's Notice about a Customer Model is handled under Section 14.5, and HIPS keeps the record of each Import and its attestation as evidence (Section 11.4).
08The Free Allowance, Credits, promotions, MTO and the Affiliate Program
One Free Allowance per person. Where HIPS grants a Free Allowance, it is granted once for each Verified Person, on the first Workspace that person owns, however many Accounts or Workspaces are involved, and only after that person has passed identity verification, which always includes a face check, automated or by a person (Terms of Service Sections 4.1, 4.8 and 4.9). You shall not obtain, or try to obtain, more than one Free Allowance, including through a second Account, a borrowed or false identity, an identity document of yours other than the one with which you first verified, another person acting for you, or a Workspace opened after a pause or a closure. HIPS recognizes the same Verified Person across Accounts and verifications by the identifier of a natural person that Terms of Service Section 4.2 describes, which is never derived from images of the face or any other biometric data and which withdrawing consent to the face check does not delete. That identifier recognizes the same identity document, and a document of the same country that carries the same personal identification number.
Credits are not for trade. Credits are a prepayment for HIPS's own Services (Common Clauses Section 3.7). You shall not sell, buy, transfer, pledge or exchange Credits, the Free Allowance or Promotional Credits, or offer them as a reward or a payment to anyone, except as the Terms of Service allow.
Payments. You shall not:
- (a)pay with a card or an account you are not authorized to use, including through a payment that an agent acting with an API Key of your Workspace makes with the card of a person who has not given you authority to use it;
- (b)test cards or payment details against the Services;
- (c)dispute or reverse a payment that you know was authorized, for Services that were supplied as the Agreement provides; or
- (d)pay with funds derived from crime, or from a transaction that breaches Sanctions Laws.
This Section does not limit any right that the rules of a card network or Applicable Law give you to dispute a payment.
Promotions. A promotion, and the Promotional Credits granted for it, are for the Verified Person and the Workspace to which HIPS grants them, once, under the terms HIPS states for the promotion. You shall not obtain a promotion or its Promotional Credits more than once, or through an Account, a Workspace or a card that is not entitled to it.
Self-dealing. You shall not, alone or with others, arrange for the Requests of a Workspace that you, or a person acting with you, belong to or pay for to be served by a Device of yours or of a person acting with you, in order to turn the Free Allowance, a promotion, a discount or Credits into Earnings or Commissions. HIPS does not place a Workspace's work on a Device of a Verified Person who belongs to that Workspace, and you shall not try to defeat that rule.
MTO and Crypto-Assets. Common Clauses Section 3.12 applies to everyone this Policy binds: HIPS does not encourage, invite, recommend or advise anyone to acquire, hold, sell or trade MTO or any other Crypto-Asset, Crypto-Assets carry extreme risk, including the total loss of their value, and acquiring or holding one is each person's own decision. HIPS accepts no payment and makes no payout in MTO, USDC or any other Crypto-Asset, connects no wallet and offers no benefit for holding one (Common Clauses Section 3.11). HIPS pays every Provider and Program Partner only in money, to a payout account in the payee's own name (Common Clauses Section 3.11(e)). You shall not:
- (a)offer a Crypto-Asset, a wallet or a balance of MTO to HIPS as a means of payment or as a ground for any benefit;
- (b)connect, or try to connect, a wallet to the Services;
- (c)misrepresent your Country, your residence or your location, or use a Workspace, a Member or a payout account in another country, to obtain a payment option, a benefit or a payout that is not offered where you are; or
- (d)obtain a payment option, a benefit or a payout for, or offer one to, a person to whom HIPS does not offer it, including a US User or any person in the United States, to whom HIPS never offers MTO in any form (Common Clauses Section 3.11(b)).
Whether or not HIPS ever offers a Crypto-Asset feature, you shall not present anything you publish or send in connection with the Affiliate Program or HIPS's name as an encouragement, invitation, recommendation or advice from HIPS to acquire, hold, sell or trade MTO or any other Crypto-Asset.
The Affiliate Program. The Affiliate Program opens only when HIPS decides to open it and announces it at hips.com/affiliates, and the Affiliate Program Agreement governs it. This Section binds everyone who uses a Referral Link or a Referral Code, from their first use, whether or not they have accepted the Affiliate Program Agreement. For Program Partners, Sections 6, 12, 15 and 16 of the Affiliate Program Agreement set out the same rules in full, and a breach of this Section in the Affiliate Program is a breach of that agreement. While the Affiliate Program is open, you shall not:
- (a)refer yourself, including through a second Account, a second card, a relative or a person acting with you, or refer a person who is in substance yourself or who acts with you or for you;
- (b)arrange with a Referred Person, or with anyone else, a sign-up, a payment, usage, Earnings or a payout that serves to create a Referral or a Commission rather than a genuine use of the Services, or refer Devices that you control or whose payouts go to a destination you control;
- (c)create Referrals with false, borrowed, purchased or synthetic identities, with sign-ups or registrations that no genuine person intends to use, or with visits produced by automated means, click farms, or paid or incentivized traffic designed to look like genuine interest;
- (d)record or claim a Referral without the Referred Person's deliberate visit through your Referral Link or entry of your Referral Code, including by forcing or planting a referral cookie, on hips.com or on the Console, adding your Referral Code or a `ref` parameter to a link that the person did not choose to follow from your promotion, hidden pages or frames, automatic redirects, pop-ups, browser extensions or other software, or guessing Referral Codes;
- (e)promote a Referral Link or a Referral Code without disclosing, clearly and conspicuously and close to it, that you earn a Commission from HIPS, as Section 15.2 of the Affiliate Program Agreement requires;
- (f)promote it by communications sent without the consent that Applicable Law requires, in bulk, by automated means or where the rules of the place prohibit it, or in HIPS's own support, documentation, community or social media channels;
- (g)make any statement about HIPS, the Services, its prices, the Free Allowance, the Referral Discount, availability, capacity, security or the use of data other than what HIPS publishes; state, estimate or imply what anyone earns or can earn by supplying GPU time to HIPS or from the Affiliate Program, or present either as an income scheme or a business opportunity; or publish a fake review or testimonial;
- (h)in anything connected with the Affiliate Program or HIPS's name, mention MTO, USDC or any other Crypto-Asset, encourage, invite, recommend or advise anyone to acquire, hold, sell or trade one, link to any way of acquiring one, or describe HIPS, Credits, Commissions or Earnings as an investment or a source of yield or return (Common Clauses Section 3.12);
- (i)buy, bid on or use "HIPS", "hips.com", "HIPS Payment Group", "Hyperscale Inference Processing Systems", "HiP", the names of HIPS's products and services, or any variation, misspelling, abbreviation, translation or combination of them, as a keyword or targeting criterion, or in the text or address of a paid advertisement, in any search engine, social network, app store, marketplace or other advertising service;
- (j)use any of the names (i) lists, or anything confusingly similar, in a domain name, social media handle, app, extension or package name, business name or email address, or publish a page, app or account that a reasonable person could take for HIPS's own;
- (k)present yourself as HIPS, as acting for HIPS, or as HIPS's employee, agent, reseller or official partner, present the Affiliate Program as connected with HIPS Payments, or use HIPS's name, logo or other marks except as the Affiliate Program Agreement allows;
- (l)sell, publish or share Invitation Codes (Section 4.4), or state or imply that a Referral Link or a Referral Code gives anyone an invitation, priority or a place on the early-access list;
- (m)create an Account, register for early access or pay for another person in order to make a Referral, or submit another person's details to HIPS; or
- (n)share a Commission with the people you refer, offer them money, Credits, gifts or anything else of value to sign up, pay or provide GPU time, other than the Referral Discount that HIPS gives, or publish a Referral Link or Referral Code on a coupon, voucher, cashback or deal website, unless HIPS has agreed to it in writing.
What follows from a breach of this Section. What a breach of this Section 8 means for Credits, the Free Allowance, discounts, promotions, the Referral Discount, Referrals, Commissions and Earnings is governed by the Terms of Service, the Provider Agreement and the Affiliate Program Agreement, including the rule that a Referral obtained in breach of this Section earns no Commission (Affiliate Program Agreement Section 6.6). Section 12 sets out HIPS's Measures.
09Rules for Providers
Important for Providers: honest supply, and Verification Work you cannot tell apart
HIPS checks the work of every Device by measurement, including Verification Work that is not marked and cannot be told apart from other work. What the Provider App reports is advisory; eligibility, Earnings and standing rest on what HIPS measures or verifies (Section 11.6). This Section forbids tampering with the Provider App, misreporting a Device, running more than one Device on one GPU, trying to recognize Verification Work, and inspecting what your Device processes. Earnings do not accrue for work that is not performed as the Provider Agreement requires, and, where HIPS establishes fraud after review, no Earnings accrue for the Devices concerned for the period in which it occurred and the Provider pays HIPS's documented costs of the fraud (the Fraud Costs), as the Provider Agreement provides (Common Clauses Section 3.6(d) and (e)).
Who supplies: Providers. You supply GPU Time only:
- (a)through your own provider account. A natural person holds only one provider account, and an organization holds only one provider account, which it holds through the Account of the one natural person who represents it; a natural person who represents an organization's provider account holds no provider account of their own (Provider Agreement Section 3.8), and HIPS verifies an organization, its representative's authority and its owners as Provider Agreement Sections 3.7 and 3.8 describe;
- (b)from Devices you own, or that you are entitled to use for this purpose with the consent of every person whose consent is needed, such as an employer, a landlord or the person who pays for the electricity; and
- (c)for your own account, never for or on behalf of another person, and never through another person's provider account, except as the representative of an organization under (a), acting through that organization's provider account and for its account.
An operator of data-center capacity supplies the Approved Pool only under its own agreement with HIPS.
The Provider App as HIPS ships it: Providers. You shall run only the Provider App and the engines HIPS signs and pins, unmodified, with their integrity checks, their sandbox and their updates working. You shall not:
- (a)modify, patch, hook, inject code into, instrument or debug the Provider App, its engines or its sandbox, or run a debug build or a build HIPS has not released;
- (b)run an engine outside its sandbox, replace an engine, or put another program in its place;
- (c)operate a Device through a script or a program that speaks the device protocol in place of the Provider App;
- (d)disable, bypass or falsify an integrity check, a signature or digest verification, the engine sandbox or an update;
- (e)intercept, read or alter the traffic between the Provider App and its engines, or between the Provider App and HIPS; or
- (f)decompile, disassemble or otherwise reverse engineer the Provider App.
This Section does not restrict what Applicable Law permits you to do with software you are entitled to use and does not allow to be excluded by contract, such as observing, studying or testing its functioning, and it does not limit the rights an open-source license gives you in an engine's own code. It governs how you take part as a Provider.
Honest work: Providers. You shall not:
- (a)run a model, checkpoint, quantization, precision or engine other than the one assigned, or report one in place of another;
- (b)pass work to another machine, a remote GPU, a hosted service or another person;
- (c)return cached, replayed, fabricated, padded, truncated or altered results;
- (d)report usage, timings or outcomes that the Device did not produce, or a completion for work that was rejected, canceled, not assigned or not run; or
- (e)accept work you do not intend to perform, or stall work you have accepted.
Honest reporting: Providers. You shall not falsify or manipulate anything the Provider App or a Device reports, including the hardware, the GPU's identity and identifiers, memory, capabilities, operating system, driver, app version, thermal and power state, idle state, availability, location, Country and network, including the round trips that a Device measures to HIPS's regional processing locations. You shall not use a virtual private network, a proxy or any other means to disguise a Device's location or your Country (Common Clauses Section 7.6). A Device that HIPS locates in a Restricted Jurisdiction takes no work; a Device that connects through an anonymizing network, or whose location HIPS cannot confirm, takes no Request of a Buyer until HIPS has confirmed its location, and the Provider App may reach HIPS outside a virtual private network, where the Device's operating system allows it, so that HIPS can confirm it; and a Device takes Requests of Buyers only while it is located in a Device Country, as Provider Agreement Section 3.3 defines it (Common Clauses Sections 3.2(j) and 7.6). When a Device first connects from a country that is not yet a Device Country, HIPS assesses that country for transfers of Request Data and opens it as a Device Country where the assessment allows; until then, the Device takes only test work, which is paid as other work is wherever paid work is open in the Provider's Country (Provider Agreement Sections 1.10 and 4.9). You shall not conceal that a Device runs in a virtual machine, or use virtualization to present hardware that a Device does not have.
One GPU, one Device: Providers. You shall not:
- (a)enroll one GPU or one machine as more than one Device, or run more than one instance of the Provider App, or more than one enrolled Device, on one GPU;
- (b)enroll a GPU or a machine that is enrolled as another Device connected at the same time, whether yours or another Provider's; a GPU moved from one machine to another, including one you bought from another person, may be enrolled in the machine that now holds it, and HIPS releases its earlier record on its own while the other machine is offline (Provider Agreement Section 6.3);
- (c)clone, image or copy an enrolled machine, an enrolled virtual machine, a device key or a device certificate, or move a device key or certificate to another machine; or
- (d)enroll a decoy Device, that is a Device with no GPU, or no real capacity, behind what it reports.
Verification Work: Providers. HIPS measures every Device with Verification Work. You shall not:
- (a)try to detect, identify, distinguish, log, collect or predict Verification Work, including jobs whose result HIPS knows, probes, sampled jobs and decoy models;
- (b)treat or run Verification Work differently from other work;
- (c)keep, share, publish or exchange information about Verification Work, its inputs or its results with anyone, including other Providers; or
- (d)use what you learn of Verification Work to pass HIPS's checks without performing work as assigned.
Nothing in this Section 9.6 or in Section 9.7 prevents you from keeping, unaltered, the local log and the support bundle that the Provider App itself writes, which hold metadata such as timestamps, job and attempt identifiers, state changes and engine events and no Request Data, or from using them in a request for review under Section 13.6, a complaint or a Dispute, and giving them for that purpose to your legal advisers and to the body that hears it. Using them to detect, distinguish or predict Verification Work remains a breach of (a).
What your Device processes: Providers. While a job runs, your Device holds the model and the content of the job in memory, and the engine that ran it keeps that content in its memory after the job until a later job overwrites it or the engine stops (Common Clauses Section 3.14(c)(v)). You shall not:
- (a)access, inspect, read from memory, copy, record, log, store, extract, reverse engineer, disclose or use any model, Input or Output that your Device processes, or use any of them to train, fine-tune or evaluate any model, other than as the Provider App does in running the job;
- (b)try to identify a Buyer, a Workspace, an End User or any other person from what your Device processes, or to learn which Customer Model or Workspace a job belongs to;
- (c)write a Customer Model, an Input or an Output to disk, keep it after the job, or use a download link outside the Provider App or from another network address; or
- (d)modify the Provider App, its sandbox or its engines to reach any of them.
When your Device processes content, you process it only on HIPS's instructions and only to run the job, under the processor terms of the Provider Agreement.
Customer Models on Devices: Providers. At launch no Customer Model runs on a Device. If HIPS later lets Devices run Customer Models, a Device receives them only after you turn that option on for it, and a breach of Section 9.7 concerning a Customer Model is a serious breach of this Policy.
Availability and assignment: Providers. You shall not manipulate idle detection, your schedule, your availability, the drain on your return, cancellations or rejections in order to obtain work, a ranking or a share of work you would not otherwise receive. HIPS decides which work, if any, to assign to each Device, and you have no right to receive work (the Provider Agreement).
Collusion: Providers. You shall not share a GPU, a result cache or a relay with another Provider, arrange with a Buyer for work to be steered to your Devices, or take part in self-dealing under Section 8.5.
Device authorization and sign-in: Providers. You shall not:
- (a)persuade or trick another person into approving a device or a tool, or into entering a device authorization code that you started;
- (b)enter a device authorization code that a device or tool under your control did not show you;
- (c)guess, collect or relay device authorization codes; or
- (d)copy or use a Tool Token outside the tool to which HIPS issued it.
You shall revoke an authorization from your account page as soon as you suspect that it is misused, and tell HIPS.
The device connection: Providers. You shall not send malformed, oversized or flooding messages to the service that Devices connect to at providers.hips.com, or test its protocol, except under Section 10.
Payouts: Providers. You shall receive Earnings only to a payout account in your own name, in your Country, and only by a payout method offered there (Common Clauses Sections 3.11(e) and 11.1). You shall not give false payout, tax or residence information, use another person's payout account, or let another person receive your payouts.
Safe and lawful operation: Providers. You shall operate your Devices safely and lawfully. You shall not disable or bypass the thermal, power or other safety protections of a Device's hardware or operating system in order to run work. You remain responsible for the electricity, the network, the premises and the permissions your Devices use, as the Provider Agreement provides.
Caches and learning features: Providers. HIPS decides which caches, and which learning, logging and telemetry features, of the Provider App and its engines are on. The prompt cache between jobs is off on every Device until HIPS turns it on under Terms of Service Section 9.9(e), and until then no Device keeps a prompt cache or reuses anything of one job for another (Terms of Service Section 9.9(e); Common Clauses Section 3.14(c)). You shall not:
- (a)turn on, re-enable, configure or emulate a prompt cache, any other cache of the state an engine computes for an Input, or any other cache between jobs, or any learning, fine-tuning, adaptation, logging or telemetry feature of an engine, that HIPS has turned off or has not turned on;
- (b)keep any part of a job, including its Input, a prefix of it or the state an engine computed for it, for use in another job, or answer one job with state kept from another; or
- (c)report input read from a cache, or a result served from a cache, as work the Device performed, or try to be paid for cached input that HIPS has not turned on for the Device.
Work performed in breach of this Section is not performed as the Provider Agreement requires, and a breach of (a) or (b) is also a breach of Section 9.7. Declining the Provider App's crash reports, or turning them off, as Provider Agreement Section 5.14 allows, is not a breach of this Policy.
10Security and security testing
What is prohibited. Except under Section 10.3, you shall not:
- (a)probe, scan, fuzz or test the vulnerability of the Services, or of any system, Device or network connected to them;
- (b)breach or circumvent any authentication, authorization, encryption, sandbox, limit, quota, metering or other security measure;
- (c)access, or try to access, data, Accounts, Workspaces, Customer Models, Hub Tokens, Devices or systems you are not authorized to access;
- (d)submit Inputs, files, models or messages crafted to exploit, crash, exhaust or take control of the Services, an engine, a decoder, the Import, a worker, a Device or the service that Devices connect to, including decompression bombs and malformed files;
- (e)use the Services to attack, probe or fingerprint a Provider's Device or an Approved Provider's infrastructure;
- (f)interfere with the Services by a denial-of-service attack, flooding or any other means; or
- (g)phish, socially engineer or impersonate HIPS's staff, a Buyer or a Provider to obtain access or information.
What is not Security Research. Measuring the latency, throughput and cost of your own Requests within your limits, load testing your own application within the limits of your Plan, and evaluating a model's behavior through ordinary Requests are not Security Research. Section 5 still applies to the content of those Requests, unless HIPS agrees otherwise in writing.
Authorized testing. You may carry out Security Research on the Services only with HIPS's prior written authorization, within the scope, the period and the conditions that the authorization states, and in compliance with Section 10.4. HIPS decides whether to give an authorization.
Rules for authorized testing. Even with an authorization, you shall:
- (a)test only the systems the authorization names, and never a Provider's Device, an Approved Provider's infrastructure, a payment provider, an identity verification provider, a model hub or any other third party's systems;
- (b)use only Accounts and Workspaces that you own, reach no data of others beyond the minimum needed to show a vulnerability, and stop at once and tell HIPS if you reach any;
- (c)not degrade the Services for others, including by denial-of-service or high-volume automated testing;
- (d)not use social engineering, phishing or physical access;
- (e)not keep, copy or disclose data you reach, and delete it when HIPS asks or the testing ends;
- (f)report every vulnerability you find to HIPS promptly, with what HIPS needs to reproduce it; and
- (g)keep each vulnerability confidential until HIPS has fixed it or agrees to its disclosure, and in any case until ninety (90) days after you reported it, or, if HIPS releases a fix within those ninety (90) days, until thirty (30) days after the fix is released, unless you and HIPS agree a longer or shorter period in writing.
Reporting a vulnerability. If you find a vulnerability in the Services or in the Provider App, the `hi` command-line tool, an SDK or another tool HIPS supplies, report it to security@hips.com, which HIPS publishes at /.well-known/security.txt on hips.com together with its Vulnerability Disclosure Policy at https://hips.com/security. HIPS acknowledges each report without undue delay, triages and tracks it, and tells you when it is fixed. Research that follows the Vulnerability Disclosure Policy is Security Research that HIPS authorizes under Section 10.3, and HIPS will not bring a claim against you, or ask any authority to act against you, for research that follows it in good faith. That authorization covers only systems HIPS controls, and never a Provider's Device, an Approved Provider's infrastructure, a host of HIPS Payments or any other third party's systems. HIPS pays no reward for a report.
Credentials you find. If you come across an API Key, a Tool Token, a Hub Token or another credential for the Services that is not yours, you shall not use it, and you shall report it to HIPS.
The hosts of the Services. The Services are provided under hips.com on the hosts that Terms of Service Section 6.14 lists, and on no other host under hips.com: ai.hips.com, for the API, its batch interface and the device sign-in of tools, which HIPS answers in the European Union at eu.ai.hips.com and ai.hips.com relays; eu.ai.hips.com, us.ai.hips.com and jp.ai.hips.com, the regional addresses of the API; console.hips.com, for the Console; providers.hips.com and the hosts under it of HIPS's regional processing locations, for the endpoints to which the Provider App connects; and weights.ai.hips.com, for the downloads of Devices. The HIPS website and its documentation are at hips.com; www.hips.com answers every request with a permanent redirect to hips.com before any cookie is set, except a request for the script address of a service worker that an earlier website used, which it answers with a script that unregisters itself and a Clear-Site-Data header limited to storage and cache (Terms of Service Section 6.14(b)). The HIPS website, its documentation and www.hips.com are not part of the Services, but this Section 10 applies to them as it applies to the Services. Every other host under hips.com, including every host that HIPS Payments uses, is not part of the Services and is run by legal entities other than HIPS and the Operating Company, and HIPS carries on no part of the HIPS Payments business (Common Clauses Section 3.10): this Policy does not govern it, HIPS cannot authorize any Security Research on it under Section 10.3, and nothing in this Policy permits you to test it. You shall send API Keys only to ai.hips.com or a regional address of the API, and Tool Tokens only to ai.hips.com, a regional address of the API or, for the Provider App enrolling a Device or renewing its certificate, providers.hips.com, and enter your HIPS sign-in only on console.hips.com, and you shall report to HIPS any page on another host that presents itself as the Services.
11What HIPS inspects and when
Request content is not read, kept or logged. HIPS never trains on, retains or logs Prompt Data (Common Clauses Section 3.14). HIPS does not review the content of Requests to detect breaches of this Policy, except through the automated safeguards of Section 11.10, which read in memory only and keep only the record of a refusal, and as Section 11.3 provides. The only content HIPS holds is what Common Clauses Section 3.14(c) lists: what a Buyer asks HIPS to hold in order to deliver what it asks for, for the periods the Terms of Service publish, which are Batch Files, the results of task Requests and jobs, uploaded files, Knowledge documents and files kept for later use, and, once HIPS offers agents, an agent's own definition and instructions and its kept disk (the input file of a Batch is deleted as soon as its Batch ends, and its output and error files thirty (30) days after they are produced, unless the Buyer sets a shorter expiry or deletes them sooner, as Terms of Service Section 6.22 provides); and content that Applicable Law requires HIPS to keep, report or disclose, only for that purpose. A prompt cache is held only in the GPU memory of the data-center workers that HIPS runs, never on disk or other storage, and on a Device only as Terms of Service Section 9.9(e) provides, kept apart for each Workspace and made unreadable no later than the longest life that Terms of Service Section 9.9 sets, and no person at HIPS reads it (Terms of Service Section 9.9). To verify Cached Input Tokens and to keep prompt caches apart, HIPS keeps the Prefix Ledger, which holds, for each cache scope of a Workspace, keyed hashes (HMAC, under a key that HIPS keeps secret) of blocks of the tokens of the Inputs of its Requests to language models served by data-center workers whose prompt cache is on, and no prompt text (Terms of Service Section 13.7). Terms of Service Section 17.4 describes what a data-center operator could access.
Metadata, rate signals and reports. HIPS detects possible breaches from the metadata it keeps for every Request, Account and Workspace, without reading content: for example usage, the models and pools used, outcomes and errors, rates and bursts, the Country and network class of client addresses, payment events and disputes, links between Accounts that resolve to the same Verified Person, Import refusals and their reasons, and the other signals the Privacy Notice lists. HIPS also acts on Notices and on reports from Buyers, Providers, payment providers, model hubs, rights holders and authorities.
When a person at HIPS may access content. A person at HIPS may read Prompt Data that HIPS holds at a Buyer's request, and the files of a Customer Model, only where Applicable Law requires it (Common Clauses Section 3.14(f)), including where Applicable Law requires HIPS to decide a Notice about content it stores (Section 14.4). Each access needs the approval of a second administrator, is granted for a limited time, and is recorded in an audit log. HIPS's automated checks of a Customer Model at Import run without a person reading its files, and verification re-runs are compared automatically, without a person seeing them. HIPS holds no content of a Request for review only because it was flagged, and a person at HIPS never reviews content for an abuse review, a security incident or a debugging request. HIPS keeps the content of a Request that the safeguards of Section 11.10 flag only where Applicable Law requires it to preserve or report that content, and only as long as that law requires.
Customer Models and Hub Tokens. At Import, HIPS's systems check every file automatically, including its format, its structure, code, pickle content, operators, external data, templates and the license the hub states; they re-save and sign the files HIPS accepts; and HIPS profiles the model on its own capacity before any deployment. HIPS keeps a record of each Import, including the hub, the repository, the pinned revision, the license record, the hub account the Hub Token signs in as where the hub states it, the Hub Token's last four characters and the attestation, and may use it as evidence under this Policy. Nobody at HIPS reads a stored Hub Token. While a security incident or a rights holder's Notice concerning a version is open, HIPS keeps its copy of that version for that purpose only, even if you delete the version, and tells you that a hold applies.
No training. HIPS makes no training copy of any Request, on any Plan, and never uses Prompt Data to train, fine-tune, distill or evaluate any model (Common Clauses Section 3.14(a)), so no Request is screened, masked, curated or kept for that purpose, and no Measure concerns a Workspace's contributions to Training.
Measurement of Devices. HIPS measures every Device, including by Verification Work, by comparing its timing with what its class of hardware achieves, by probes of its memory and capacity, by running sampled jobs again, in memory only, on a GPU that HIPS itself operates and comparing the results, by counting the usage that passes through HIPS's systems, by hardware identifiers, which HIPS keeps as the Provider App reports them and compares with those of other Devices and with its hardware blocklist, as Privacy Notice Section 3.9 describes, and by signs of virtualization. Data the Provider App sends is advisory; eligibility, Earnings and standing rest on what HIPS measures or verifies. The content of a sampled job is held only in memory for its comparison, never written to disk or other storage, and is gone the moment the comparison ends (Common Clauses Section 3.14(c)(iii)). HIPS does not tell Providers which jobs are Verification Work.
Automated decisions and human review. HIPS's systems may take some Measures on measured signals without a person deciding each case, such as flags, temporary limits, raised verification, holds of Earnings pending review, and the suspension of a Device whose results fail HIPS's checks. HIPS's systems apply automatically only these Measures: for a Workspace or an Account, a flag and a temporary limit; for a Device, a flag, more Verification Work, a hold of its Earnings pending review under Provider Agreement Section 11, and its suspension where a byte-exact mismatch of its results coincides with a second strong signal, a probe of its memory or capacity fails, or its GPU is enrolled as more than one Device connected at the same time. A possible match against a sanctions list holds onboarding, a payment or a payout until a person at HIPS has reviewed it. HIPS does not disclose the thresholds (Section 3.6). A person at HIPS decides the suspension or closure of an Account or a provider account, a forfeiture, a claim for Fraud Costs and a report to an authority. You may ask a person to review any Measure under Section 13.6, and the Privacy Notice describes your rights about automated decisions.
Content classifiers. HIPS classifies the content of Requests automatically for the models that Section 11.10 covers. HIPS does not classify the content of other Requests to detect breaches of this Policy. Every automated classifier reads content in memory only and keeps nothing of it but the record of a refusal, its category and its time (Common Clauses Section 3.14). Before HIPS starts classifying other Requests, it changes this Policy, the Terms of Service and the Privacy Notice with notice under Common Clauses Section 9.2, and the change applies only to Requests HIPS accepts after it takes effect.
Personal data. HIPS processes personal data for the purposes of this Section 11, and of Sections 12 to 14, as the Privacy Notice describes.
Safeguards for models that generate images, video or audio. HIPS lists and serves a catalog model that can generate or manipulate realistic images, video, audio or similar material depicting people only with reasonable and adequate technical safety measures and other safeguards that reliably prevent it from generating or manipulating the material that Sections 5.2 and 5.3 describe, including the automated classification of that model's Inputs and Outputs. HIPS refuses the Requests, and withholds the Outputs, that those safeguards flag (Section 13.5), and Section 5.2 applies to what they detect. HIPS may apply the same safeguards to a Customer Model of that kind, and Section 7.8 applies to Customer Models. These safeguards are not a general monitoring of content, and Section 3.4 applies to them. Before HIPS lists such a model, it tests the safeguards against a documented set of Inputs designed to produce that material, including foreseeable attempts to circumvent them, and lists the model only if they reliably prevent it; it tests them again after every change of the model, its engine or the safeguards, and corrects observed or reported misuse, as Article 5(1a) of the AI Act requires where it applies. HIPS keeps no content the safeguards flag, only the record that it refused the Request or withheld its Output, with the category and the time, except as Section 11.3 provides.
12Enforcement
When HIPS acts. HIPS may take a Measure, with immediate effect and without prior notice, where it has reasonable grounds to believe or suspect that:
- (a)you, a Member, an End User, or anyone using your Account, a Workspace, your credentials or your Devices has breached this Policy;
- (b)a Workload, a Customer Model or any content is Illegal Content or breaches this Policy;
- (c)a use of the Services creates a security risk, or a risk of liability for HIPS, its Buyers, its Providers or others; or
- (d)Applicable Law, an order of a competent authority or Sanctions Laws require it.
Sections 12.6 and 12.7 add rules for Consumers and for US Users.
The Measures. The Measures HIPS may take, alone or together, are:
| Measure | What it does | Applies to |
|---|---|---|
| Flag and review | Records the signals and opens a review; nothing changes for you | Buyers, Providers and Program Partners |
| Refuse | Refuses or cancels a Request, a Batch, an Import, a deployment, a feature, a payment method or a payout method, or refuses a Referral or a Referral Code | Buyers, Providers and Program Partners |
| Limit | Lowers limits for a period, including the limits of Batches, caps or revokes API Keys and Tool Tokens, restricts models, pools or features, or withholds a promotion or a Referral Discount pending review | Buyers |
| Remove a Customer Model | Stops a version's deployments, disables access to it, deletes HIPS's copy once the evidence HIPS needs is preserved, and deletes the related Hub Token | Buyers |
| Pause a Workspace | Pauses inference for the Workspace, with the effects the Terms of Service describe | Buyers |
| Raise verification or stop work | Increases Verification Work, narrows a Device's eligibility, or stops assigning work to a Device | Providers |
| Hold | Holds Earnings, Commissions or payouts, or holds a Referral for review, as and for as long as the Provider Agreement or the Affiliate Program Agreement allows | Providers and Program Partners |
| Suspend | Suspends a Device and revokes its certificate, suspends a provider account or a Program Partner's participation in the Affiliate Program, or suspends a Verified Person, which blocks sign-in for every Account of that person, drains their Devices and holds their payouts as the Provider Agreement and the Affiliate Program Agreement allow (Section 12.8) | Buyers, Providers and Program Partners |
| Close | Closes an Account, a Workspace or a provider account and revokes its credentials and Device certificates, or ends a participation in the Affiliate Program, under the Terms of Service, the Provider Agreement or the Affiliate Program Agreement | Buyers, Providers and Program Partners |
| Preserve | Keeps the records and Customer Model copies HIPS lawfully holds that are relevant to the breach, for as long as that purpose needs, and Prompt Data only as Section 12.11 allows | Buyers, Providers and Program Partners |
| Report | Reports to the competent authorities, and to others as Section 12.9 allows | Buyers, Providers and Program Partners |
Scope of a Measure. A Measure may cover the whole or part of the Services; one Request, model, Customer Model version, API Key, Tool Token, Workspace, Device, Account, provider account or participation in the Affiliate Program; or, where HIPS has reasonable grounds to believe that the breach or the risk extends to them, every Account, Workspace, provider account, participation in the Affiliate Program and Device of the same Verified Person.
Duration. A Measure lasts for the period HIPS states, or until its ground ends or a review lifts it. HIPS lifts or narrows a Measure when it finds that its ground no longer applies.
Stopping work is not a Measure against your rights. HIPS's decision to assign no work, or less work, to a Device in the ordinary course of placement is not a Measure. Where HIPS stops assigning work to a Device because of a suspected breach of this Policy, it is a Measure, and Section 13 applies.
Closure: Consumers. If you are a Consumer:
- (a)HIPS takes a Measure against you only in proportion to its ground;
- (b)HIPS closes your Account or a Workspace with immediate effect only on the serious grounds that Terms of Service Section 24.2 states (fraud HIPS has established after review, a serious or repeated breach, an amount unpaid after notice, the application of Sanctions Laws, or a requirement of Applicable Law or a competent authority), and terminates the Provider Agreement, or closes your provider account, with immediate effect only on the serious grounds that Provider Agreement Section 17.5 states, and terminates the Affiliate Program Agreement with immediate effect only on the serious grounds that Affiliate Program Agreement Section 19.6 states; and
- (c)in any other case, HIPS ends the Agreement only as the Terms of Service, the Provider Agreement or the Affiliate Program Agreement allow it to end without cause, on the notice they require.
Termination at will: US Users. Nothing in this Policy limits HIPS's right under the Terms of Service, the Provider Agreement or the Affiliate Program Agreement to suspend or terminate the Account, a Workspace, the provider account or the participation in the Affiliate Program of a US User at any time, for any reason or no reason, as those documents provide.
Money. What a Measure means for Credits, Fees, Earnings, Commissions and payouts, including holds, the non-accrual of Earnings for work not performed as required, the forfeiture of Earnings and the Fraud Costs on fraud established after review, the netting of HIPS's documented costs, and the treatment of Credits when a Workspace is closed, is governed by the Terms of Service, the Provider Agreement, the Affiliate Program Agreement and Common Clauses Sections 3.6 and 3.13. Your liability for a breach of this Policy is not limited by Common Clauses Sections 5.3 to 5.8 (Common Clauses Section 5.9(e)), and Common Clauses Section 6 applies to it.
Reports and orders of authorities. HIPS may report a breach of this Policy, and give the information needed to identify the person concerned, to the competent authorities, and, where Applicable Law permits, to rights holders, payment providers, model hubs and others harmed by it, disclosing no more of your Inputs, Outputs or Customer Models than shows the breach; such a report is a disclosure that the confidentiality clauses of the Agreement permit (Terms of Service Section 21.7). Where HIPS becomes aware of information giving rise to a suspicion that a criminal offense involving a threat to the life or safety of a person has taken place, is taking place or is likely to take place, it informs the competent law enforcement or judicial authorities, as Article 18 of the Digital Services Act requires. HIPS complies with orders of competent authorities to act against Illegal Content or to provide information, and tells you of such an order and of the effect given to it where Applicable Law, including Articles 9 and 10 of the Digital Services Act, requires.
Identity record. When HIPS closes an Account, a Workspace or a provider account, or ends a participation in the Affiliate Program, for fraud, abuse or a reason under Sanctions Laws, it keeps the reference of the verification and the identifier of a natural person that Terms of Service Section 4.2 describes, a keyed hash computed from identity document data and never from images of the face or any other biometric data, whether or not the person's consent to the face check stands, so that a later invitation, verification or participation of the same person can be recognized and refused, even under a new verification, for five (5) years after the closure, as the Privacy Notice describes. That identifier recognizes the same identity document, and a document of the same country that carries the same personal identification number.
Preservation. HIPS may preserve the records and Customer Model copies it holds that are relevant to a breach, a Notice, a Dispute, a security incident or a request of a competent authority, for as long as that purpose needs. It preserves Prompt Data that it holds at a Buyer's request only where Applicable Law, or a binding order of a competent authority, requires it, only that Prompt Data and for no longer than the law or the order requires (Common Clauses Section 3.14(c)(iv)). In each case HIPS keeps what it preserves apart from other data, gives access to it only as Section 11.3 allows and suspends its deletion for that purpose only, as Terms of Service Section 13.8 and the Privacy Notice describe.
Court relief. HIPS may apply to any competent court for interim relief to stop a breach of this Policy (Common Clauses Sections 13.6 and 14.3(b)).
13Statement of reasons and review
When HIPS gives reasons. HIPS gives a Statement of Reasons, on a Durable Medium, for every Measure, except a flag or a review that does not change what you can do.
Measures based on content. Where a Measure is taken on the ground that information you provided, such as an Input, a Customer Model or a file, is Illegal Content or breaches this Policy, HIPS gives the Statement of Reasons at the latest when the Measure takes effect, as Article 17 of the Digital Services Act requires. This Section does not apply where HIPS does not have your electronic contact details; and, where the Measure gives effect to an order of a competent authority under Article 9 of the Digital Services Act, HIPS gives the Statement of Reasons at the time that order allows (Article 9(5) of the Digital Services Act).
Other Measures. For a Measure taken on another ground, such as fraud, payment fraud, the integrity of a Device's work, a security risk or Sanctions Laws, HIPS gives the Statement of Reasons when the Measure takes effect, unless Applicable Law prohibits it or telling you would prejudice the prevention or investigation of fraud, of a security incident or of a breach of Sanctions Laws; HIPS then gives it when that reason ceases.
What a Statement of Reasons contains. A Statement of Reasons states, clearly and specifically:
- (a)the Measure, what it covers and how long it lasts;
- (b)the facts and circumstances relied on, in general terms where more detail would disclose the matters in Section 3.6, or the identity of a Notifier that Applicable Law protects;
- (c)whether the Measure follows a Notice, HIPS's own detection or an order of an authority;
- (d)whether the Measure was taken, or the facts detected, by automated means;
- (e)where the Measure rests on Illegal Content, the legal ground relied on, and where it rests on a breach of the Agreement, the provision concerned and why the conduct breaches it; and
- (f)how to ask for a review, and the other means of redress available to you (Section 13.7).
For a Measure under Section 13.3, a short statement of the Measure, the category of its ground, the provision concerned and the route to review may serve as the Statement of Reasons.
Automated refusals. When HIPS's systems refuse an Import or a Request and show the reason, with a reference to this Section 13, at the moment of the refusal, that display is the Statement of Reasons for the refusal.
Asking for a review. You may ask HIPS to review a Measure, free of charge, within six (6) months after HIPS gives you the Statement of Reasons for it, through Review a decision in the Console or, if you cannot sign in to the Console, by email to legal@hips.com, quoting the reference that the Statement of Reasons gives. Your request should say why you consider the Measure wrong and give the information that supports it. A person at HIPS who has the authority to change the Measure and who, where practicable, did not take it, reviews it in a timely, diligent, non-discriminatory and objective manner, and tells you the outcome and its reasons on a Durable Medium. Where the review finds that the ground for the Measure did not apply, HIPS reverses the Measure without undue delay. A request for review does not suspend a Measure unless HIPS decides that it should. A Provider's dispute of an adjustment, a Hold or a forfeiture of Earnings follows the Provider Agreement, and a Program Partner's dispute of a Hold, a reversal or a forfeiture of Commissions follows the Affiliate Program Agreement.
Other redress. A review does not limit your other means of redress: a Dispute on your Track under Common Clauses Sections 12 to 15, including, if you are a Consumer, the courts of your home country and the out-of-court resolution Common Clauses Section 15.6 describes; and a complaint to a competent authority.
14Reporting abuse and illegal content
Anyone may report. Any individual or entity, whether or not it uses the Services, may notify HIPS of content in the Services that it considers to be Illegal Content or a breach of this Policy, such as a Customer Model, an uploaded file or a job result that HIPS stores, and of misuse of the Services by a Buyer or a Provider.
How to report. Notices are submitted electronically through HIPS's notice form at https://hips.com/legal/report or by email to abuse@hips.com, both listed on the Legal Information Page. A Notice is an operational message, not a notice under Common Clauses Section 10.3.
What a Notice should contain. To let HIPS act on it, a Notice should contain:
- (a)a sufficiently substantiated explanation of why the Notifier considers the content to be Illegal Content or a breach of this Policy;
- (b)a clear indication of where the content is, such as the model hub repository and revision of a Customer Model, a request identifier, or any other information that lets HIPS identify it;
- (c)the Notifier's name and email address, except in a Notice concerning a child sexual abuse offense, which may be anonymous; and
- (d)a statement that the Notifier believes in good faith that the information and allegations in the Notice are accurate and complete.
What HIPS does with a Notice. HIPS confirms receipt of a Notice without undue delay where the Notifier gave an email address, and decides on every Notice in a timely, diligent, non-arbitrary and objective manner. HIPS tells the Notifier of its decision without undue delay, with the means of redress available, and says whether it used automated means to process the Notice or to decide on it. Where a Notice concerns content that HIPS stores at a Buyer's request, a person at HIPS examines that content only where Applicable Law requires it in order to decide the Notice, such as Article 16 of the Digital Services Act, with the approval and the record of Section 11.3, and HIPS may disable access to the content concerned without reading it. Where HIPS takes a Measure, it gives the person concerned a Statement of Reasons under Section 13. HIPS does not disclose a Notifier's identity to the person concerned, except where it is needed to deal with the Notice, such as a Notice by a rights holder about its own rights, or where Applicable Law requires it.
Rights holders and Customer Models. A rights holder who considers that a Customer Model infringes its rights, or breaches the license under which it was imported, may notify HIPS under Section 14.2. On a Notice that gives HIPS reasonable grounds to believe so, HIPS may stop the model's deployments, disable access to the version, pause the Workspace while it reviews the Notice, delete its copy once it has preserved the evidence it needs, and delete the related Hub Token, and it uses the record of the Import and the attestation as evidence. HIPS cannot act on the source repository or revoke a Hub Token at its hub; where that is needed, it tells the Notifier to ask the hub, or the token's holder, to do so. Section 14.10 also applies to a notification of claimed copyright infringement under the law of the United States.
Reports by Buyers and Providers. You shall tell HIPS promptly:
- (a)of a breach of this Policy that you become aware of in the use of your Account, your Workspaces or your Devices, including by a Member or an End User;
- (b)if you believe that an Account, a credential, a tool or a Device of yours has been compromised; and
- (c)if you are a Provider, if the Provider App or an engine behaves in a way that may expose content or harm your machine.
Notices made in bad faith. You shall not submit a Notice that you know to be unfounded, or submit Notices in order to harass a user or to disrupt the Services. HIPS decides on every Notice, but where it has warned a Notifier who frequently submits manifestly unfounded Notices, it may decide on that Notifier's later Notices in a summary procedure. A user who abuses the notice mechanism is in breach of this Policy.
Emergencies. If a person's life or safety is in immediate danger, contact the local emergency services first. HIPS's notice form is not an emergency service.
Authorities. Competent authorities contact HIPS through its single point of contact for authorities, published on the Legal Information Page with the languages it accepts (Common Clauses Section 3.2(b)).
United States copyright notifications. HIPS's agent designated under 17 U.S.C. 512(c)(2) to receive notifications of claimed copyright infringement in material stored at the direction of a user, such as a Customer Model, an uploaded file or a Batch File, is: Copyright Agent, HIPS Payment Group Limited, 5 The Courtyard Building, Carmanhall Road, Sandyford, Dublin 18, Ireland; email legal@hips.com; telephone as given in HIPS's designation in the directory of the U.S. Copyright Office. The designation is registered with the U.S. Copyright Office before HIPS first stores material at the direction of a user, and is listed on the Legal Information Page. A notification is sent to that agent in writing and contains the elements of 17 U.S.C. 512(c)(3), and HIPS also treats it as a Notice under this Section 14. On a notification that substantially complies with 17 U.S.C. 512(c)(3), HIPS expeditiously removes or disables access to the material concerned and takes reasonable steps promptly to tell the Buyer. A Buyer whose material is removed or disabled may send the designated agent a counter-notification that meets 17 U.S.C. 512(g)(3). HIPS then sends a copy to the person who sent the notification and restores the material, or access to it, as 17 U.S.C. 512(g)(2) provides, unless that person first tells the designated agent that it has filed an action seeking a court order to restrain the Buyer from the infringing activity. A person who knowingly and materially misrepresents that material is infringing, or that it was removed or disabled by mistake, may be liable under 17 U.S.C. 512(f). HIPS terminates, in appropriate circumstances, the Workspaces and Accounts of Buyers who are repeat infringers.
15Contact
Where to write. Notices, reports and requests under this Policy go to the addresses below, which the Legal Information Page also lists.
- Notices of Illegal Content and abuse reports
- the notice form at https://hips.com/legal/report, or abuse@hips.com
- Requests to review a Measure
- Review a decision in the Console, or, if you cannot sign in, legal@hips.com (Section 13.6)
- Notifications of claimed copyright infringement under 17 U.S.C. 512
- the designated agent named in Section 14.10
- Security vulnerabilities
- security@hips.com, also published at /.well-known/security.txt on hips.com, under the Vulnerability Disclosure Policy at https://hips.com/security
- Single points of contact for authorities and for users
- the Legal Information Page
- Legal notices under the Agreement
- as Common Clauses Section 10.3 provides
- Postal address
- HIPS Payment Group Limited, 5 The Courtyard Building, Carmanhall Road, Sandyford, Dublin 18, Ireland