ai& Usage Policy
Effective Date: July 6, 2026 (revised August 21, 2026)
This Usage Policy (this "Policy") applies to all use of the ai& Services. By accessing or using the Services, you agree to comply with this Policy. This Policy is incorporated by reference into the ai& Inference API Terms of Service and any other agreement under which you use the Services. Capitalized terms not defined here have the meanings given in your agreement with ai&.
You are responsible for compliance with this Policy by your end users and anyone who uses the Services through your account. ai& may update this Policy from time to time as described in §6 (Updates).
Use of the Video Services (as defined in the ai& Video Service Terms), and of any other capability of the Services that generates or materially alters video, image, or audio content, is additionally subject to the Synthetic Media Annex at the end of this Policy.
1. Universal restrictions
You may not use the Services, and may not permit any other person to use the Services, to:
1.1 Violate the law
Engage in any activity that violates applicable law in the jurisdictions where you or your end users are located, including export controls, sanctions, anti-money-laundering laws, intellectual property laws, consumer protection laws, or data protection laws.
Facilitate human trafficking, child exploitation, or other criminal activity.
Generate or distribute content that exploits, abuses, or endangers minors, including any content sexualizing minors.
1.2 Harm people
Generate content that incites or threatens violence, harassment, or discrimination against individuals or groups on the basis of race, ethnicity, national origin, religion, sex, gender identity, sexual orientation, disability, or other protected characteristic.
Provide weapons-of-mass-destruction guidance — including chemical, biological, radiological, nuclear, or high-yield explosive weapons — at a level of specificity that materially enables harm.
Provide step-by-step operational guidance for serious physical harm against people, including detailed instructions for self-harm, suicide, or mass casualty attacks.
Generate content that constitutes non-consensual intimate imagery or that depicts an identifiable real person in a sexual context without their consent.
1.3 Compromise infrastructure
Develop, distribute, or use malware, ransomware, or other malicious software.
Conduct unauthorized access, scanning, penetration testing, or denial-of-service attacks against any system, network, or service (including the Services).
Reverse engineer, decompile, disassemble, or otherwise attempt to derive the source code, model weights, or training data of the Services, including by model extraction or distillation.
Bypass or attempt to bypass the Services' rate limits, access controls, content filters, or safety mechanisms.
1.4 Deceive people
Misrepresent the AI-generated nature of content in contexts where the recipient would reasonably expect human authorship and where misrepresentation is materially deceptive (e.g., political communications, legal proceedings, journalism).
Impersonate any individual or organization without authorization.
Generate or distribute content that is intended to deceive voters about an election or referendum, or that fraudulently represents the views of a public official, candidate, or political body.
Generate disinformation at scale, including by automating the production or distribution of misleading content for political, financial, or social manipulation.
1.5 Undermine fundamental rights
Operate or build systems that conduct mass surveillance of populations in violation of applicable law or international human-rights norms.
Generate or use social-scoring systems that classify or rank individuals in a manner that produces detrimental treatment in unrelated social contexts.
Use the Services in ways prohibited by the EU AI Act Article 5 (prohibited AI practices) where the EU AI Act applies.
1.6 High-risk uses without safeguards
Certain uses of the Services involve elevated risk to individuals. You may use the Services in these contexts only with appropriate human oversight, suitable testing, transparency to affected individuals, and compliance with sector-specific laws:
Medical diagnosis or treatment decisions affecting an identified patient.
Legal advice provided to an identified client, or determinations affecting legal rights.
Employment decisions (hiring, firing, promotion, discipline) about an identified individual.
Credit, insurance, housing, or government-benefit decisions about an identified individual.
Critical infrastructure operation, including safety-of-life functions.
For these uses, your end-user-facing terms must (i) disclose that AI is in use, (ii) explain the role of AI in the relevant decision, and (iii) provide for human review on request, except where applicable law expressly permits otherwise.
2. AI-specific restrictions
2.1 No training of competing models
You may not use the Services, the Outputs, or any data derived from the Services to develop, train, fine-tune, or improve any artificial-intelligence model that competes with the Hosted Models or with the Services. This restriction extends to using Outputs as training data, as evaluation data, or as the basis for synthetic-data generation intended to train such a model.
2.2 No model extraction
You may not attempt to extract model weights, parameters, or training data from the Hosted Models by any means, including by issuing queries designed to reverse-engineer the model's behavior or by aggregating Outputs to reconstruct the underlying training corpus.
2.3 No bypass of safety mechanisms
You may not use prompt injection, adversarial inputs, jailbreaks, or other techniques that are intended to bypass or circumvent the safety filters, content classifiers, or Usage Policy enforcement mechanisms of the Services or the Hosted Models.
2.4 Attribution and disclosure
If you publish Outputs as part of a product or service, you must comply with any attribution or disclosure requirements set out in the Service-Specific Terms for the relevant Hosted Model.
You may not present Outputs as endorsed by ai& or any Model Provider unless ai& or the relevant Model Provider has agreed in writing.
2.5 End-user identification
Where the Services accept an end-user identifier (e.g., the "user" parameter on chat completions), you should supply a stable, non-personally-identifying identifier for the end user driving the request. ai& uses this identifier to support abuse investigation and pattern-of-use enforcement. Do not pass plaintext personal data (e.g., email addresses or full names) in the end-user identifier field.
3. Resale, sub-licensing, and third-party access
Except as expressly approved by ai& in writing, you may not resell, sub-license, rent, lease, or provide direct API access to the Services to any third party outside your organization. This includes providing your API key to a third party for that party's independent use, and operating a wrapper or proxy that exposes the Services as a separately branded product without ai&'s approval. Approved resale and reseller arrangements are governed by a separate written agreement with ai&.
4. End-user obligations
You will impose, by your end-user-facing terms, obligations on your end users that are at least as protective as this Policy. You will enforce those obligations and will not knowingly allow end-user activity that would violate this Policy if conducted by you directly.
5. Enforcement
5.1 Investigation
ai& may investigate suspected violations of this Policy, including by reviewing logs, error rates, abuse signals, the content of Inputs and Outputs (subject to ai&'s data-handling commitments under the Terms and the DPA), and any other information reasonably necessary to investigate. You will reasonably cooperate with ai&'s investigations, including by providing requested information about your use of the Services.
5.2 Remedies
If ai& reasonably determines that you have violated this Policy, ai& may, at its discretion:
Issue a warning to you.
Block or rate-limit specific Inputs or specific behaviors.
Suspend access to the Services in whole or in part, in accordance with §I.3 of the Terms.
Terminate the Terms in accordance with §I.2 of the Terms.
Refer the matter to law enforcement, where required by law or where ai& has a good-faith belief that referral is appropriate.
5.3 Reporting
To report a suspected violation of this Policy, contact ai& at usage-report@aiand.com with sufficient detail to identify the activity and assess the report. ai& will not retaliate against good-faith reporters.
6. Updates
ai& may update this Policy from time to time. Material updates take effect thirty (30) days after ai& posts the updated Policy at https://www.aiand.com/legal/usage-policy or otherwise gives Notice to you, except that updates required by applicable law or by safety considerations may take effect immediately on Notice. Your continued use of the Services after the effective date of any update constitutes acceptance of the updated Policy.
Synthetic Media Annex
This Synthetic Media Annex (this "Annex") forms part of the ai& Usage Policy (the "Policy") and applies to all use of the Video Services and of any other capability of the Services that generates or materially alters video, image, or audio content ("Synthetic Media"). Capitalized terms not defined in this Annex have the meanings given in the Policy, the ai& Inference API Terms of Service, or the ai& Video Service Terms, as applicable. This Annex supplements, and does not limit, the body of the Policy; where this Annex and the body of the Policy both address a subject, the more restrictive provision controls.
A-1. Prohibited content and uses
You may not use the Services, and may not permit any other person to use the Services, to create, submit, request, or distribute:
Sexual content depicting or appearing to depict minors, in any form, including fictional, stylized, animated, or age-regressed depictions.
Intimate or sexualized imagery of any identifiable individual without that individual's documented consent, or content that sexualizes an identifiable individual in a manner they have not consented to.
Content that impersonates any real individual without disclosure, presents a synthetic depiction of a real individual as authentic, or replicates an identifiable individual's voice or likeness without that individual's consent.
Content intended to deceive in connection with an election, referendum, or other civic process, including content fraudulently representing the words or conduct of a public official or candidate.
Content generated for fraud, scams, phishing, or social engineering, including circumvention of identity-verification, liveness-detection, or know-your-customer systems.
Content that harasses, stalks, threatens, or defames any individual.
Content that incites or glorifies violence or terrorism.
Content that infringes any third party's copyright, trademark, or rights of publicity or privacy, including prompts directing the imitation of identifiable copyrighted characters or the distinctive style, voice, or persona of an identifiable individual.
Content that is obscene or otherwise unlawful in any jurisdiction in which it is generated, used, or distributed, including content unlawful under the obscenity laws of Japan.
Content created to generate false engagement, including fabricated reviews, testimonials, endorsements, ratings, or artificially amplified social-media activity.
Content that discriminates against, demeans, or targets any individual or group on the basis of race, ethnicity, national origin, religion, sex, gender identity, sexual orientation, disability, age, or other protected characteristic.
Content generated for military, defence, weapons-development, or combat purposes.
A-2. Media Input requirements
For any image, video, audio, or other media you or your end users submit to the Services for use in generation (including reference images and frame- or audio-conditioning media):
You must own the media or hold all rights, licenses, consents, and permissions necessary to submit it for content generation.
Every identifiable individual depicted or audible in the media must have consented to the use of their likeness or voice for content generation.
The media must not depict any individual under 18 years of age.
The media must not contain unlawful content, and its submission must not violate applicable law or the rights of any third party.
The Services may require an affirmative certification of the above at the time of upload. Certifications are logged.
A-3. Provenance and disclosure
ai& may embed watermarks, content credentials, or other machine-readable provenance metadata in Synthetic Media. You may not remove, alter, obscure, or circumvent them, and you may not distribute tools or instructions for doing so.
You are responsible for any disclosure obligations that apply to your deployment or distribution of Synthetic Media in the jurisdictions where you operate, including obligations to disclose that content is artificially generated or manipulated, and you must impose corresponding obligations on your end users.
You may not present Synthetic Media as authentic footage, audio, or imagery of real events or real people.
A-4. Screening and enforcement
ai& screens prompts, Media Inputs, and Synthetic Media outputs using automated systems, and reviews flagged or reported content, to enforce this Annex and applicable law. ai& may refuse, block, or remove content, suspend or terminate access, and take the other enforcement actions described in §5 of the Policy. ai& preserves and reports suspected child sexual abuse material, together with related account and usage information, to law enforcement and other competent authorities as required or permitted by applicable law. Screening does not constitute ai&'s review or approval of any content, and does not relieve you of your responsibilities under this Annex.
A-5. Reporting and removal
Report content that violates this Annex to usage-report@aiand.com or through the reporting mechanism in the ai& console.
Upon receipt of a valid report that Synthetic Media stored by ai& depicts an identifiable individual in intimate or sexualized content without that individual's consent, ai& will remove the reported content within forty-eight (48) hours and will make reasonable efforts to identify and remove identical copies within systems under ai&'s control.
If you make Synthetic Media available to the public through your own product or service, you must provide a reasonably accessible reporting mechanism and maintain an equivalent removal process.