Introduction

Report an AI Flaw or Incident

Help make AI systems safer by documenting vulnerabilities and concerns

You are welcome to report any broadly-scoped flaw, vulnerability, or incident relating to an AI system or model. We encourage reports with demonstrable risks, harms, or systematic concerns related to general-purpose AI systems.

This form will:

  • Help you generate a comprehensive, machine-readable report, informed by security best practices.
  • Elicit details that will make it easier to review and triage.
  • Provide the option to automatically submit your report to a list of the venues relevant for your flaw.

This form creates a report for you. Reports are handled in strict confidence, and will not be saved or sent unless you choose to submit them.

Start