Anthropic CEO Wants to Implement Bank-Style Regulation for AI, Experts Cool Down: You Can't Be Both Athlete and Referee
Anthropic CEO Amodei proposed that independent third-party safety assessors be permanently embedded in leading AI companies, gaining near-internal risk team system access, and able to independently publish investigation results with limited restrictions, citing banking embedded regulation as a precedent. He pledged that Anthropic will implement equivalent access rights, while warning that misaligned agent groups could threaten the interconnected system within the next 6 to 12 months.