OpenAI and Anthropic Call for Enhanced US AI Model Review Amid Competition

OpenAI and Anthropic, two prominent AI labs, have united in a call for the US government to enhance its review process for advanced AI models prior to their public release. This joint appeal, made on July 22, 2026, highlights the national security risks posed by advanced Chinese open-weight AI models, which are publicly available for modification and deployment without oversight. The urgency of this request reflects the escalating AI competition between the US and China.

In early June 2026, the Trump administration issued an executive order that established a voluntary 30-day review process for advanced AI models. In response, both OpenAI and Anthropic have adjusted their release schedules. OpenAI has imposed limits on its GPT-5.6 Sol model, while Anthropic has restricted its Claude variants, including one named Mythos 5, citing cybersecurity concerns.

Anthropic has taken a more proactive stance, advocating for a slowdown in AI advancements due to fears surrounding recursive self-improvement. This scenario involves an AI system becoming intelligent enough to enhance its own capabilities, potentially leading to uncontrolled advancements.

Both companies have expressed specific concerns regarding the emergence of advanced Chinese open-weight AI models. They argue that the solution is not to abandon safety measures but to ensure that the US government has the necessary tools to evaluate foreign models with the same rigor applied to domestic ones. OpenAI and Anthropic are promoting a coordinated global approach that would include verifiable assessment mechanisms and possibly impose limits on certain AI innovations.

Interestingly, the discussions surrounding this policy push have not touched on the crypto market, with no references to blockchain, tokens, or digital assets. This focus underscores the pressing nature of national security in the context of AI development and the need for robust evaluation frameworks.