In the United States, the Trump administration is weighing a set of administrative actions aimed at bolstering the oversight of state-of-the-art artificial intelligence models, in light of potential national security threats. Sources from the tech sector and policy consultants, who wished to remain anonymous, revealed that the White House has been deliberating for a while over the creation of a formal 'review-and-approval' framework for sophisticated AI models. Under this framework, corporations would be mandated to present their novel models to the government for assessment prior to public release, and would require official clearance before deployment. The assessment would primarily target risks including cybersecurity vulnerabilities and the potential for military exploitation. The policy change was directly spurred by the capabilities demonstrated by Anthropic's Mythos model, which showcased formidable cyberattack prowess.