Ro Khanna’s alternative approach to U.S.-China AI safety collaboration · Digg

Ro Khanna’s alternative approach to U.S.-China AI safety collaboration

Axios reports that Khanna is spelling out his approach in a letter to Secretary of State Marco Rubio.

1 Source, 1h ago, first seen 1h ago

An illustrated portrait of Rep. Ro Khanna surrounded by ballot elements.

TLDR

Axios reports that Ro Khanna is offering an alternative U.S.-China AI playbook. His letter to Secretary of State Marco Rubio outlines how he would tackle AI safety collaboration with China.

Featured Source

Rep. Ro Khanna is proposing a binding U.S.-China agreement on advanced AI safety, arguing that the two countries should treat loss-of-control risks more like nuclear arms control than ordinary technology competition.

Axios reports that the California Democrat set out the approach in a letter to Secretary of State Marco Rubio. Khanna is the ranking member of the House Select Committee on China and has been positioning AI safety as a major policy issue.

Pauses, audits and a verification system

Khanna calls for a global pause on recursive self-improvement, in which AI systems design and test improvements to themselves, along with a separate pause on superintelligence. He also wants common protocols for kill switches and mandatory pre-release model audits.

The proposal tries to address the central objection to bilateral safety promises: verification. Khanna recommends technical verification tools backed by an international inspection agency or nongovernmental organization that both sides could trust.

His letter arrived as President Donald Trump and Chinese President Xi Jinping discussed AI without producing a major safety agreement. Khanna argues that Washington and Beijing share an interest in preventing humans from losing control of advanced systems, even when the countries disagree on most other strategic questions. The plan remains a congressional proposal, not U.S. policy or a negotiated treaty.

1 Source

@axiosExclusive: Ro Khanna offers alternative U.S.-China AI playbook https://www.axios.com/2026/09/25/ro-khanna-alternative-us-china-ai-playbook?utm_campaign=mrf-utm_source=x&utm_medium=organic_social&utm_campaign=editorial&utm_source=twitter&utm_medium=social&mrfcid=202609256ab536242db635028385f9a7

Combined views

8.3K

1 Source, first seen 1h ago

5 likes

Ro Khanna’s alternative approach to U.S.-China AI safety collaboration

TLDR

Pauses, audits and a verification system

Ro Khanna’s alternative approach to U.S.-China AI safety collaboration

TLDR

Pauses, audits and a verification system

1 Source

Combined views

Pauses, audits and a verification system

Pauses, audits and a verification system

Combined views

Sentiment

Summary

Sentiment

Summary

Sentiment

Summary

1 Source