The arrival of a new generation of powerful artificial intelligence models, like Anthropic’s Mythos, has begun to crack the White House’s hard-line stance on promoting the technology, as top officials confront security risks posed by tools that can easily find flaws long buried in computer code.
