AI News
The company asked Congress for capability-based requirements while supporting four California bills and an incident-reporting framework.
OpenAI published a policy statement asking Congress to create mandatory, capability-based national safety requirements for advanced artificial-intelligence systems. The company said federal action should be paired with continued state activity until Congress establishes a national framework. This account draws on Axios and distinguishes confirmed events from statements, allegations, proposals, or preliminary findings.
Why it matters: A frontier developer asking for enforceable rules narrows the debate from whether regulation is needed to what triggers, tests and disclosures it should require. The proposal remains a company position until lawmakers convert it into binding text.
Read full articleAI News
Jacob Coxon said laboratories are prioritizing competitive development over safety; Anthropic did not treat his prediction as an independently established forecast.
AI researcher Jacob Coxon resigned from Anthropic and published a statement warning that frontier laboratories are prioritizing competitive model development over safety. Coxon said he had spent about three years doing pretraining research at Anthropic and OpenAI, giving him direct experience in the organizations he criticized. This account draws on Associated Press and Axios and distinguishes confirmed events from statements, allegations, proposals, or preliminary findings.
Why it matters: The resignation is notable because it converts private risk concerns into a public personnel decision by someone who worked on frontier model training. Its value lies in the governance questions it raises, not in treating one researcher’s forecast as settled science.
Read full articleAI News
The dispute over extracting capabilities from frontier models moved into diplomacy ahead of planned talks between Presidents Trump and Xi.
China rejected U.S. claims that Chinese artificial-intelligence developers were engaged in aggressive, malicious and industrial-scale efforts to extract capabilities from American frontier models. The dispute concerns model distillation, a broad family of techniques in which outputs from one system can be used to train or improve another model. This account draws on Associated Press and distinguishes confirmed events from statements, allegations, proposals, or preliminary findings.
Why it matters: Distillation disputes sit at the boundary of cybersecurity, intellectual property and national-security policy. How governments define evidence and permitted model access could reshape cloud controls, export rules and research collaboration.
Read full articleAI News
The alignment researcher and federal AI-standards adviser will serve on the nonprofit board and observe the company’s public-benefit corporation board without a vote.
OpenAI appointed alignment researcher Paul Christiano to the OpenAI Foundation Board and to the board’s Safety and Security Committee. Christiano will also serve as a non-voting observer on the OpenAI Group public-benefit corporation board, a role that provides access without a formal vote. This account draws on Axios and distinguishes confirmed events from statements, allegations, proposals, or preliminary findings.
Why it matters: The appointment places a prominent technical safety researcher closer to formal oversight. The practical test will be what information, authority and dissent mechanisms the board and committee use when commercial and safety priorities conflict.
Read full article