Sriram Krishnan, the former senior White House AI policy adviser, appeared on CNBC on July 1, 2026, to discuss Anthropic’s Claude Mythos, stating it is effective at discovering cyber weaknesses and that critical systems need protection. The statement coincided with the lifting of US export controls on Anthropic’s Mythos 5 and Fable 5 variant, which had been imposed in mid-June 2026 after reports of a jailbreak. The model was leaked internally in late March 2026, and Anthropic officially launched Claude Mythos Preview on April 7, 2026. Through a limited initiative called Project Glasswing, the company granted early access to roughly 50 trusted partners, including banks and security firms, where Mythos uncovered over 10,000 critical software vulnerabilities. Before departing the administration, Krishnan helped establish a 30-day government testing window for frontier AI models. Krishnan highlighted that Mythos is both a potential defensive cybersecurity tool and a threat vector, as the same capability that finds vulnerabilities could be used to exploit them.