They urged leaders to enact permanent testing for frontier models to prevent opacity as competing models from abroad become more available.

There are proposals, distinct from the safety framework, to create an independent AI regulatory agency modeled after financial regulators to oversee safety reviews.

Discussions are underway about an autonomous regulatory body for AI safety oversight, a concept supported by industry figures.

The White House signals a non-mandatory approach to evaluating open-weight AI models, telling developers that voluntary safety tests will not apply to open-weight systems like Nvidia's Nemotron and Meta's Llama.

Open models such as Nemotron and Llama have publicly accessible core components, while proprietary closed models remain controlled by companies like OpenAI, Google, and Anthropic.

The administration plans to exempt open-weight models from voluntary safety testing, drawing a distinction with closed models under tighter corporate control.

The push follows warnings from Anthropic about vulnerabilities and past testing breaches by major firms, fueling a sense of urgency for U.S. oversight.

The policy timing aligns with a forthcoming high-profile summit between Trump and Xi Jinping, underscoring AI leadership as a strategic battleground between the U.S. and China.

The Washington meeting comes as the U.S. and China pursue leadership in AI, with the Trump-Xi summit casting a spotlight on this tech rivalry.

China’s Moonshot AI released the Kimi K3 model, heightening concerns about U.S. AI leadership and spurring investments in data center infrastructure.

Five Democratic senators urged Congress and the administration to establish permanent, mandatory testing for frontier models, warning against opaque, case-by-case restrictions while Chinese AI remains cheaper and more accessible.

The same lawmakers pressed for enduring testing for the most advanced U.S.-made models, arguing that policy should not leave frontier AI unreviewed.