On Tuesday, Anthropic announced that its Claude AI discovered new weak spots in two encryption systems. Claude found a hidden pattern in HAWK, a digital signature designed to resist future quantum computers, in 60 hours after experts had reviewed it for two years. According to Anthropic, cracking the smallest HAWK key fell from 2^64 steps to 2^38, roughly 67 million times less work, and the fix requires doubling the key size. For AES, which scrambles most internet traffic, Anthropic attacked a weaker version with 7 of the usual 10 scrambling rounds; Claude invented a shortcut called the Möbius Bridge that made the best known attack 200 to 800 times faster. Claude worked alone for three days, and two humans spent nearly a month checking its answer. Anthropic says humans cannot keep pace with the model. These attacks do not affect Bitcoin or Ethereum, which rely on a different system. Google plans to switch by 2029 under its migration timeline, and US agencies face a 2031 quantum deadline. NIST must now weigh a public attack on a scheme it was still considering. Anthropic hinted at more unreleased results and plans a workshop with academics in the coming weeks.