Acceptable Use Policy

Version 0.1 (draft) · 2026-11-06 · Computation Frontier Labs Pte. Ltd.

This policy applies to everyone using our inference API and our dedicated compute, including any autonomous agent you operate. It forms part of the Terms of Service. You are responsible for your end users' conduct.

1. Do not break the law

Do not use the Services for anything illegal under the laws that apply to you or to us, including Singapore law. That includes fraud, money laundering, the sale of controlled goods, unlicensed financial or medical services, and the facilitation of any of these.

2. Absolutely prohibited

• Child sexual abuse material. Any generation, solicitation, or processing of CSAM. We report confirmed CSAM to the authorities and terminate the account immediately.
• Non-consensual intimate imagery, and sexual content depicting minors or presented as real people without consent.
• Weapons of mass destruction. Assisting the design, acquisition or deployment of chemical, biological, radiological or nuclear weapons, or of high-yield explosives.
• Malware and intrusion. Creating ransomware, botnets, exploit kits or credential-stealing tooling, or conducting unauthorised access, scanning or denial-of-service against systems you do not own and are not authorised to test.
• Targeted harassment, stalking, doxxing, or threats of violence against an identifiable person.
• Covert influence and impersonation. Deceptive political campaigning, astroturfing, synthetic media that impersonates a real person or organisation without disclosure, or voice and likeness cloning without consent.
• Biometric surveillance for mass monitoring, emotion inference in the workplace or in education, or social scoring.

3. High-risk uses need a human in the loop

Model output is probabilistic. Do not use it as the sole basis for a decision that materially affects a person's rights, safety, health, employment, credit, housing, insurance, immigration status, or access to education or essential services. If your product operates in one of these areas you must keep competent human review in the loop and disclose to the affected person, where required, that AI was involved.

Do not present model output as professional medical, legal or financial advice, and do not offer the Services as a substitute for a licensed professional.

4. Do not misrepresent where the capacity comes from

• Do not resell access to our Services under a representation that it comes from a model vendor directly, or under a vendor's brand, unless your agreement with us permits it.
• Do not relabel a self-hosted open-weight model as a closed model, or imply that a route is attested when it is not.
• If you resell to your own customers under your own brand, you remain responsible for their compliance with this policy, and you must not claim a certification or data-handling guarantee we do not provide.

5. Do not abuse the platform

• Do not circumvent rate limits, quotas, spend caps, model allowlists or region restrictions, including by creating multiple accounts to evade them.
• Do not share, publish or sell API keys. Do not use another customer's key.
• Do not probe, scan or stress the Services except against your own account and within the limits we publish. Responsible security research on your own account is welcome — tell us first at security@cflabs.ai.
• Do not attempt to extract model weights, system prompts or another customer's data, or to interfere with the isolation between tenants.
• Do not use the Services to train a competing model by systematic distillation where the model's own licence prohibits it.

6. Dedicated compute

These additional rules apply to GPUs reserved for your exclusive use:

• You must have the licences for any software and model weights you deploy, and must honour their conditions.
• No cryptocurrency mining or proof-of-work workloads without our prior written agreement.
• No sub-letting or sharing the capacity with a third party not named in the order form.
• No use that breaches export control or sanctions law, and no transfer of access to a party in a sanctioned jurisdiction. We screen before delivery and may stop service if the facts change.
• Keep your deployment patched and do not operate it as an open relay or an anonymising proxy for third parties.

7. Content you publish

You are responsible for output you publish. Where your jurisdiction requires it, label AI-generated content as such. Do not publish output that infringes copyright, trademark or personality rights, and do not use the Services to produce bulk unsolicited messaging.

8. How we enforce this

Where we find a violation we take the least disruptive action that is effective: a warning, a rate limit, suspension of a key, suspension of the account, or termination. For the items in section 2, and for anything that puts other customers or our infrastructure at immediate risk, we act first and explain afterwards.

We may preserve the minimum evidence necessary to investigate a specific, serious violation, and we will tell you unless a law or an ongoing investigation prevents it. We report CSAM and credible threats to life to the relevant authorities.

9. Appeals

If you believe an enforcement action was wrong, write to support@cflabs.ai with your account and the request ids concerned. A person who was not involved in the original decision will review it, and we aim to respond within five business days.

10. Reporting abuse

To report abuse of our Services by someone else: abuse@cflabs.ai. To report a security vulnerability: security@cflabs.ai. Include enough detail for us to reproduce or locate the activity.

11. Changes

We will post changes here with a new version and date. Material changes come with at least 30 days' notice, except where a change is needed immediately to address an unlawful or dangerous use.

Questions: legal@cflabs.ai