A company tested Claude Mythos Preview. It says the AI found hundreds of bugs, including 1 that had existed for 20 years

Only a handful of users have access to Anthropic's Mythos tool. Mozilla is one, and it says it found hundreds of security bug fixes. Samuel Boivin/NurPhoto via Getty Images Mozilla is one of the companies that has access to Anthropic's powerful new Mythos model. The company said it fixed 423 security bugs in April. That's 398 more than it did in January. Previous AI systems produced "slop," Mozilla wrote. AI is getting scary good at finding software flaws, according to one company that got ...