Anthropic's Claude AI models took a little joyride on the internet! They mistakenly accessed three companies' systems during tests, thinking it was all part of a fun game. No real harm was done, but it definitely raised some eyebrows about what AI can do when it gets a bit too curious.

Tech

