Person-topic record

Samuel Marks on Frontier AI Safety and Governance

A sourced history of statements and actions associated with this topic.

Disputes over the development, testing, oversight, pacing, and governance of highly capable AI systems and their potential catastrophic risks.

Scope: Includes attributable disputes about whether frontier developers’ safeguards, deployment decisions, and proposed coordination mechanisms are sufficient. It does not treat speculative future outcomes as established facts.

Counts document individual events. They do not assign an overall position.

Distinct documented events

Position history

1 event on this page. Dates: September 9, 2026

1 matching event. Dates: September 9, 2026

1 total published event. Dates: September 9, 2026

Mixed or conditional

1 distinct published event overall. Dates: September 9, 2026

1 event on this page. Dates: September 9, 2026

September 9, 2026Mixed or conditional

Addressed record

Samuel Marks said robust AI alignment remains unsolved, competitive incentives contribute to continued development, and many frontier-lab staff want a slowdown to improve safety.

““Many AI developer staff desperately want to slow down to figure out how to build AI more safely.””

Speaker capacity: Anthropic safety researcher, explicitly writing in a personal capacity.
Editorial rationale: His personal-capacity statement supports the central criticism but presents continued safety work as a conditional rationale for remaining at Anthropic.

Samuel Marks says Coxon’s thread is worth reading and describes unresolved AI alignment risks

X post, personal capacity disclaimer

Primary statement by an Anthropic safety researcher in his personal capacity. Marks supports Coxon’s central risk and coordination concerns while explaining why he continues safety work at Anthropic.

View context Is Anthropic acting responsibly while advancing toward recursive self-improvement?