Legal / Acceptable use
Use ML Referee responsibly.
This policy describes the rules for using ML Referee. It forms part of the Terms of Use. If this policy conflicts with the Terms, the Terms control.
1. Authorized professional use
ML Referee is for adults 18 and older performing lawful business or professional research. Use only an account and credits you are authorized to use. Do not resell access, share authentication credentials, impersonate another user, or allow an agent to exceed the authority or budget its operator granted.
2. Permitted evaluation data
Evaluation submissions may contain synthetic data, non-personal data, or genuinely anonymized data. They must not contain regulated data or identifiable personal data, including pseudonymized person-level records. Do not submit passwords, OAuth secrets, access tokens, recovery codes, payment credentials, government identifiers, malicious code, or confidential information you are not authorized to process.
Ordinary account and support contact details may be provided only through the designated account and support channels. Do not place them in evaluation datasets.
3. Illegal, harmful, or discriminatory activity
Do not use ML Referee to violate law or another person’s rights; facilitate violence, exploitation, harassment, abuse, or fraud; conduct mass surveillance; infer private or sensitive information about individuals; or support unlawful discrimination in employment, housing, credit, education, health, insurance, public benefits, or other consequential decisions.
Do not use an evaluation report as the sole basis for a decision that has legal or similarly significant effects on a person. You remain responsible for lawful domain review, human oversight, and appropriate validation of any downstream model.
4. Security and service integrity
Do not probe, scan, test, access, or alter ML Referee systems or another tenant’s information without written authorization. Do not introduce malware, overload or disrupt the service, evade authentication or isolation, bypass credits, quotes, limits, safety measures, or retention controls, or use automated traffic except through documented and authorized interfaces.
5. Deception and misuse of results
Do not use ML Referee to commit fraud, mislead others, create false evidence, or conceal material evaluation limitations. Do not claim that a report proves causation, guarantees future performance, establishes deployment readiness, or certifies a model’s safety, fairness, legality, or compliance when it does not.
You may accurately describe and share your report, but you may not imply sponsorship, endorsement, partnership, or approval by ML Referee, OpenAI, Auth0, or another provider.
6. Intellectual property and competitive misuse
Do not reverse engineer, decompile, extract, reproduce, or attempt to discover ML Referee’s nonpublic software, methodology, security controls, or internal artifacts except where applicable law expressly permits it. Do not remove proprietary notices or use the service to create a substantially similar competing service through unauthorized extraction of nonpublic material.
7. Enforcement and changes
ML Referee may reject a submission or limit, suspend, or terminate access when it reasonably believes this policy has been violated or action is necessary to protect the service, another user, or the public. Serious conduct may be reported when required by law. This policy may be updated prospectively by posting a revised effective date.
8. Questions or reports
Use the private support form to report suspected misuse or ask whether a proposed use is permitted. Do not include passwords, secrets, access tokens, or sensitive datasets in the request.