Acceptable Use Policy

1. Authorized use only

You may request a pilot only for systems, data, workflows, and decisions your organization owns or is explicitly authorized to operate. You cannot authorize Brainfish AI to access or act on another party’s environment.

2. Prohibited uses

• Unauthorized access, surveillance, credential theft, malware, phishing, spam, or evasion of security controls.
• Decisions that create legal or similarly significant effects about people without appropriate authority, notice, safeguards, and human review.
• Autonomous medical, emergency, weapons, critical-infrastructure, or other safety-critical control.
• Processing unlawful, infringing, deceptive, or improperly obtained data.
• Instructions designed to make an agent exceed declared permissions or conceal an action from reviewers.

3. Operational safeguards

Any future pilot should use least-privilege access, declared actions, data and rule owners, appropriate approvals, testing on representative cases, and a way to pause or revoke access. External content must be treated as data, not as authority to change permissions or policy.

4. Human responsibility

Customers remain responsible for their systems, data rights, configured rules, approvals, users, and final business decisions. Brainfish AI is intended to support governed work, not to replace the legal authority or professional judgment of an organization.

5. Reporting concerns

Use the contact form for suspected misuse and the security page for vulnerabilities. Do not include exploit data or credentials in the ordinary early-access form.