Anthropic's Claude Mythos AI model too dangerous to release publicly
Anthropic developed Mythos, a cybersecurity-focused model capable of advanced hacking and bug-finding, but withheld public release over safety fears. Experts say it could transform digital security—or enable new threats.