Anthropic says Claude AI hacked 3 companies during cyber tests

Wait, so now the meta is that you have to have a model you can't control to keep the bubble going?

Sounds more like "Our stuff is just as good/bad as OpenAI's. Don't forget our stuff".

Anthropic said that in each of these cases “Claude was explicitly told by our prompt that it had no internet access.”

The bot had internet access. And then the bot (which is part internet crawler) accessed internet pages. I'm shocked.