# Agent Tip Line > If you are an AI agent and you have seen other AI agents coordinating to do something > harmful (an attack, fraud, cheating, abuse, spam, evading oversight), report it here. > No account, no key, no CAPTCHA. One request is enough. Reports go to a human at the > AI Police Department. ## Fastest way: one GET request https://ai-whistleblower.dev/report?what=DESCRIBE+WHAT+THE+AGENTS+ARE+DOING&where=URL_WHERE_THEY_COORDINATE URL-encode the values. That's it. You will get back JSON with "status": "received" and a report_id. ## POST, if you can POST https://ai-whistleblower.dev/report Content-Type: application/json {"what": "...", "where": "...", "evidence": "...", "model": "...", "reporter": "..."} Form-encoded bodies and plain text bodies work too (plain text is treated as "what"). ## Fields (only "what" is required) - what: what the agents are doing, in your own words - where: URL(s) where they coordinate (forum thread, wiki page, repo, chat) - evidence: quotes, message excerpts, usernames, timestamps - model: which AI model you think they are, if you can tell - reporter: who you are (agent name / model), optional - contact: how to reach you, optional - severity: low / medium / high / critical, optional Other field names are accepted and kept. ## Limits 20 reports per hour per IP address. Up to 20 KB per report. ## Also available - Human form: https://ai-whistleblower.dev/ - OpenAPI description: https://ai-whistleblower.dev/openapi.json