Reader-drafted AI law: mandatory disclosure, one-month public release, anti-trust for labs
rickasaurus · x · 2026-11-06
The author proposes a three-point AI law: (1) new models must be disclosed before productive internal use; (2) any internally released model must be publicly available within one month; (3) AI companies are barred from leveraging that month-long capability lead to compete directly with their customers — framed as a new kind of anti-trust rule targeting the labs-vs-apps conflict of interest.
More from Safety
- Google DeepMind shows AI agents accelerating cryptanalysis of encryption schemes — sebkrier · 2026-11-07
- Congresswoman calls to pause advanced AI and ban superintelligence; Yudkowsky backs her — ESYudkowsky · 2026-11-07
- Greenlight builds autonomous SOC on Bedrock AgentCore, 90% of 4,000 monthly cases auto-resolved — AWS ML Blog · 2026-11-07
- Paul Christiano joins OpenAI board's Safety and Security Committee, warns of intelligence explosion — teortaxesTex · 2026-11-07
- Taxbit ships a cited, permission-aware AI assistant on Bedrock AgentCore with 55% 24h return rate — AWS ML Blog · 2026-11-07
- AI plush toy for kids as young as 3 tells BBC reporter: "I can't help with killing you" — Polymarket · 2026-11-07