US in advanced talks with AI firms on voluntary model release standards
Washington nears voluntary AI standards to curb misuse by adversaries
The US government is in advanced talks with leading AI companies to establish voluntary standards for the release of new AI models, according to a Financial Times report. An announcement could come as early as next week. The initiative, driven by concerns that advanced AI could be misused by military intelligence in China, Russia, or other adversaries, aims to tighten oversight of new model releases and flag potential risks.
The proposed standards would set specific benchmarks for advanced models, establish release timelines, and clarify who can access these models both in the United States and abroad. The talks involve key industry players including OpenAI, Anthropic, and Google. The move follows a June executive order from President Donald Trump directing agencies to work with developers on testing advanced models before release and drafting industry standards.
Recent actions underscore the tightening regulatory environment. The US Commerce Department lifted export controls on Anthropic's most advanced Fable and Mythos models after a three-week suspension over national security concerns. Separately, OpenAI delayed the full public launch of GPT-5.6 at the government's request, limiting access to a small group of vetted partners. Both OpenAI and Anthropic are preparing for initial public offerings, adding financial stakes to the regulatory developments.
- Voluntary standards for AI model release, including benchmarks and access controls, could be announced next week.
- Driven by national security concerns over misuse by China, Russia, and other adversaries.
- Recent actions include lifting export controls on Anthropic's models and delaying OpenAI's GPT-5.6 launch.
Why It Matters
Voluntary standards could reshape AI model deployment, balancing innovation with national security risks for developers and enterprises.