US Urged to Enhance Review of AI Models Amid Chinese Competition Fears

OpenAI and Anthropic, two leading AI research companies, have made an unexpected joint appeal to the US government. The unlikely alliance is pushing for a more comprehensive national framework for reviewing advanced AI models, particularly in light of concerns over China's rapid advancements in the field.

The joint appeal comes after both companies adjusted their release schedules in response to a voluntary 30-day review process established by the Trump administration in early June 2026. OpenAI has placed limits on its GPT-5.6 Sol model, while Anthropic has restricted access to its Claude variants, including Mythos 5.

Anthropic went further, explicitly calling for a slowdown in AI advancements due to concerns over recursive self-improvement - a scenario where an AI system gets smarter enough to make itself even more intelligent, creating a cycle of exponential growth. The company specifically flagged the emergence of advanced Chinese open-weight AI models as a concern.

Both companies are pushing for universal safety measures and a coordinated global approach that includes verifiable mechanisms for assessment and potential limits on certain AI innovations.