NY Assemblyman urges Congress to ban Recursive Self-Improvement this year, citing 2027-28 AI lab timelines
rohanpaul_ai · x · 2026-11-08
NY Assemblyman Alex Bores calls on Congress to legislate a ban on Recursive Self-Improvement (RSI) — fully automating AI research so AIs build the next AI. He cites OpenAI targeting March 2028, Anthropic employees claiming 2027, and a METR report suggesting those timelines are plausible. His core argument: any error or misalignment in current models would be multiplied across generations, and no election intervenes before labs hit that threshold. He notes the ban position enjoys rare cross-spectrum support, from Bernie Sanders and Steve Bannon to Yoshua Bengio and Arvind Narayanan.
More from AGI Musings
- Robotics olfaction may spawn new continually-learning algorithms, argues Kordem France — AnnaCiaunica · 2026-11-08
- Reuters: Meta planned 60% team cuts for AI agents, then Zuckerberg halted it — esrtweet · 2026-11-08
- Dario Amodei: 80% of Junior Lawyer Roles to Be Wiped Out in 1-5 Years — jkubicki · 2026-11-08
- The Cure That Kills an Industry: Why Progress Meets Resistance From Its Own Beneficiaries — r0ck3t23 · 2026-11-08
- The brain as the next UI: upcoming article charts BCI-meets-agentic-AI roadmap via earbuds — melnykowycz · 2026-11-08
- AI helps neurosurgery resident settle 22-year-old Crouzeix's conjecture in matrix analysis — IgorCarron · 2026-11-08