Elon Musk proposes adversarial peer reviews: major AI labs should test each other's models for safety
MikePFrank · x · 2026-09-16
On the All-In Podcast, Elon Musk floated a concrete AI safety governance idea: have major AI competitors test each other's models using their own security test harnesses, instead of "grading your own homework."
- Incentive logic: if competitors flag a model as unsafe and it later causes harm, the company would face enormous reputational and legal liability, creating natural accountability.
- He stressed any proposal must be something China is willing to accept, otherwise the West would just be handicapping itself.
- Host Jason added there's no reason these safety harnesses need to be kept private.
A rare, relatively specific governance proposal from Musk centered on adversarial cross-testing, though with no implementation details yet.
Related event: Musk Proposes Cross-Lab Pre-Release Safety Testing Among AI Companies(6 posts)→
More from Safety
- Insider: critical infrastructure firm flunked security audit, still delays basic fixes — thedealdirector · 2026-11-02
- Cyber attack on Swedish smart fridges causes mass food spoilage at supermarket — Polymarket · 2026-11-02
- Security researcher: offense is hard to automate, defending known vuln classes is easier — mboehme_ · 2026-11-02
- How to opt out of Meta, OpenAI, and Apple training AI on your data in the UK — HotMinuteUK · 2026-11-02
- AI safety researcher: AI companies entirely lack this kind of effective human oversight — davidmanheim · 2026-11-02
- offprompt: MIT-licensed MCP server hands agents API keys without ever exposing them — Vast_Ad_7338 · 2026-11-02