1 article on this topic.
Anthropic's report on detected Claude misuse describes AI agents handling reconnaissance, exploitation and data theft while humans pick targets and review output. Here is what security teams should take from it.