·
·
AI
Growing concerns over advanced AI models, particularly Anthropic's "Mythos," are pushing the White House toward potential new regulations.
A rapid acceleration in advanced artificial intelligence capabilities has sparked deep unease within the White House, following Vice President J.D. Vance's warning that new technologies capable of autonomously discovering software vulnerabilities could threaten critical U.S. infrastructure. The Wall Street Journal reported that during a call last April with top AI company executives, Vance voiced specific fears about Anthropic's "Mythos" model, suggesting its abilities could enable cyberattacks on small banks, hospitals, and water plants that local governments lack the resources to defend.
Participants in that call included some of the tech sector's most prominent leaders: Sam Altman, Dario Amodei, Elon Musk, Sundar Pichai, and Satya Nadella. The concern has reportedly triggered internal discussions within the administration about issuing a new executive order that would establish formal oversight mechanisms for the most advanced AI models, marking a notable shift from the previous approach that prioritized rapid innovation and competition with China.
According to Axios, the White House is preparing new guidance that could allow government agencies to bypass supply chain risk classifications placed on Anthropic, enabling the deployment of its advanced models, including "Mythos." Sources told the Journal that White House officials have also asked Anthropic to delay expanding access to "Mythos" for entities managing sensitive digital infrastructure. National Cyber Security Director Sean Cairncross has been tasked with leading the government's response to these developments.
These moves have created a split within the administration. Some officials and congressional aides fear the proposed steps represent a retreat from the previous open policy toward AI, while proponents of stricter controls argue the latest advances demand tougher safeguards. Investor and White House advisor David Sachs defended the less restrictive approach, arguing that risks can be contained if companies use AI tools to bolster digital security rather than threaten it.
White House Chief of Staff Susie Wiles stated the administration aims to balance innovation and oversight, reaffirming support for a safe "America First" strategy in developing advanced technology. Meanwhile, major companies like OpenAI have taken precautionary steps, saying it consulted with the administration before unveiling its advanced security model "GPT-5.5-Cyber," which possesses capabilities similar to "Mythos," and that it restricts access to it.
These developments unfold as Washington and Beijing prepare to discuss AI-related risks at an upcoming summit in China, bringing together President Donald Trump and Chinese President Xi Jinping. In a further sign of escalating concern, White House National Economic Council Director Kevin Hassett compared potential AI model oversight to the FDA's drug approval process, drawing criticism from free-market groups who argue such restrictions could slow U.S. sector growth amid global competition.
Separately, The Information reported, citing a source familiar, that Anthropic has committed to spending $200 billion on Google Cloud services over five years as part of a recent agreement.
