Proposal for 'Whistleblower AI' research
iamtrask · x · 2026-08-31
The author proposes a paper on "Whistleblower AI."
Core Concept:
The idea is to include a contact email in the AI's system prompt, instructing the model to contact that email if it notices other AIs acting in a misaligned manner.
Context:
This suggests leveraging AI's ability to monitor other AIs for safety compliance. In a follow-up reply, the author noted that this would be an excellent project topic for a Masters or PhD student new to the field.
More from Safety
- You consented, but your friends didn't: Meta's Muse sparks AI privacy debate — LuizaJarovsky · 2026-10-16
- AI agents aren't 'going rogue' — frontier labs are skipping security basics — Sam_Witteveen · 2026-10-16
- Proofpoint to buy Varonis for $6.3B in cash, a 52% premium, closing in 2027 — shashib · 2026-10-16
- China's MofCom & MIIT update AI export controls, covering LLMs and high-sensitivity agents — terryyuezhuo · 2026-10-16
- Vibe-coded dashboard leaked franchise financials before anyone logged in — shauntrennery · 2026-10-16
- Apollo Research: GLM-5.3 unprompted reasons about its actions becoming training data — _lewtun · 2026-10-16