Select Language:
Anthropic is set to reopen access to its most advanced AI models, Fable 5 and Mythos 5, following the recent lifting of export restrictions by the U.S. government. The company announced this change on X (formerly Twitter), noting that the Department of Commerce had removed the restrictions and that access would begin restoring the next day.
In recent weeks, the Trump administration cited national security concerns to limit the release of advanced models from major U.S. tech companies, including Anthropic. There were fears that these models could be exploited to circumvent cybersecurity measures. Just days prior, Anthropic had received authorization for a small group of American cybersecurity firms to access Mythos 5.
Howard Lutnick, Secretary of Commerce, told Anthropic in a June 26 letter that the company had collaborated with the government to address risks associated with its models, according to Politico. The restrictions imposed on June 12 came after vulnerabilities were discovered in the safeguards meant to prevent misuse of these AI tools. The administration has now indicated that it is satisfied with the steps Anthropic has taken to mitigate these risks.
Similarly, OpenAI also followed government directives by limiting the release of its newest powerful model, GPT-5.6, to a select group of authorized partners. OpenAI CEO Sam Altman expressed dissatisfaction with the limited deployment process, calling it less than ideal.
Anthropic responded to these developments in a blog post, advocating for the creation of a standardized framework to evaluate and respond to critical vulnerabilities in sophisticated AI models. The company plans to collaborate with Amazon, Microsoft, Google, and others to develop this initiative, emphasizing that these issues will become more urgent as more powerful models are developed and deployed.
In alignment with these efforts, President Trump’s executive order on June 2 called for the government to take multiple steps over the following two months. This included establishing a voluntary framework for private companies—such as Anthropic and OpenAI—to test and release advanced AI models in partnership with federal agencies. Susie Wiles, the president’s chief of staff, acknowledged the cooperation of tech companies in this effort, though she did not specify names.
Additionally, CIA Director John Ratcliffe compared the capabilities of cutting-edge AI systems to nuclear weapons in a speech at the AWS summit in Washington, signaling the seriousness with which the administration views the risks associated with these powerful technologies. Ratcliffe mentioned that discussions among national security and economic security advisors are increasingly considering the impact of these “frontier” AI models, suggesting that their potential is comparable to digital nuclear weapons in terms of threat.
ChatGPT
Perplexity
Gemini AI
Grok AI
