U.S. export controls are tightening around advanced AI, aiming to curb how researchers might bypass safety measures in Fable5 and to restrict international use of such models.

Anthropic is adapting to the new export guidelines for its Fable5 and Mythos5 models by restricting overseas access to ensure compliance.

Former White House AI and crypto czar David Sacks says Anthropic CEO Dario Amodei was urged to fix vulnerabilities or de-deploy the model, but declined, underscoring ongoing regulatory pressure.

OpenAI faces a statewide legal probe with a subpoena from the New York Attorney General seeking details on advertising, user engagement, data handling, protections for minors and the elderly, and internal policy practices such as model alignment.

Anthropic notes that similar vulnerabilities exist in other publicly available models, including recent OpenAI releases, and views disclosed jailbreaks as minor findings.

Regulatory scrutiny is widening from model-specific safety to the broader operation of AI services, signaling a shift in government monitoring of AI technologies.

OpenAI and Anthropic face mounting regulatory risk ahead of potential IPOs as U.S. authorities tighten oversight of AI firms.

Anthropic describes claims of bypassing Fable5 as misunderstandings and says it is working to restore services after imposing access restrictions.

A June 2 executive order signed by the President authorizes security reviews of new AI models before release, highlighting heightened national security concerns around AI.

Regulatory scrutiny of frontier AI systems is rising, with focus on export controls and cybersecurity standards in deployment.

Sam Altman postponed a planned South Korea visit amid regulatory and scheduling developments affecting OpenAI and its regional partners, including a Samsung event.

Regulators are not expected to extend Anthropic's export controls to OpenAI and other rivals, with actions likely targeted at Anthropic's models.