AOL reports that Washington reversed course on AI export restrictions, clearing Anthropic for a global push after the company patched a jailbreak pathway in its Fable 5 model, which had been blocked over national security concerns. Bloomberg strategist Neil Campling flagged the shift on July 1, noting that restrictions on Anthropic's Mythos model were eased on June 26, marking the second loosening in a week. Campling stated the fix involved "putting a new safeguard to make sure that it targets and blocks behavior that had allowed some users to bypass the security before." The policy signals a strategy of blocking China while selectively opening global markets for U.S. labs with security safeguards. AKAM surged 27% after securing a $1.8 billion Anthropic infrastructure deal. Campling warned that Google DeepMind has "gone from the leader and the most advanced model in this space to actually now playing catch up for the science industry," a problem for Alphabet as its enterprise pitch relies on DeepMind's science leadership. Dylan Patel noted Anthropic will reach 2-plus gigawatts of compute this year and roughly 10 gigawatts by the end of next year.