External Resource

When AI Builds Itself — Anthropic on Recursive Self-Improvement

Anthropic

Anthropic's June 2026 statement that AI may approach recursive self-improvement, calling for the option to slow or pause frontier development via a verifiable, internationally observed 'stop mechanism' (likened to the INF Treaty).

Theses this informs

Non-Proliferation of Dangerous AISupports
SuperintelligenceSupports
LimitationsSupports
No Intentional MisalignmentSupports
Human ResponsibilitySupports