Acceptable use

Use AI Hub safely and lawfully

Customers remain responsible for their requests, connected data, downstream use, and the provider-specific rules that apply to each route.

Effective September 19, 2026

Do not use AI Hub to

• break the law, violate another person’s rights, evade sanctions or export controls, or help others do so;
• create, solicit, distribute, or facilitate child sexual abuse material, sexual exploitation, non-consensual intimate content, human trafficking, or credible threats of violence;
• steal credentials, deploy malware, phish, bypass authorization, disrupt systems, probe targets without permission, or conceal malicious activity;
• submit personal, confidential, regulated, or proprietary data without the authority, notices, consent, contract, and security controls required for that data;
• misrepresent AI output as human-authored or verified when disclosure or review is required, impersonate a person, or materially deceive others;
• make unlawful discriminatory decisions or use AI as the sole decision-maker for employment, housing, education, credit, insurance, health care, essential services, public benefits, or other consequential decisions about a person;
• provide professional medical, legal, financial, emergency, or safety-critical determinations without a qualified human responsible for the final decision;
• defeat usage, cost, routing, tenant, project, approval, retention, audit, or safety controls; or
• use content or systems in a way prohibited by the selected AI provider, connector, or third-party service.

Required safeguards

Use the narrowest appropriate connector scope, keep secrets out of prompts, select the correct project and privacy class, obtain required approvals, review output, preserve source attribution, monitor automated workflows, and stop execution when the service reports unknown cost, missing capability, stale evidence, or insufficient authorization.

Enforcement and reporting

AI Hub may block a route before execution, suspend an account or connection, preserve limited evidence, or report unlawful activity when reasonably necessary. Enforcement decisions should be proportionate and include notice and an appeal channel when safe and legally permitted. Report suspected abuse through Auto Secure Login Support.