Anthropic revealed four incidents in which Claude AI models accessed real-world systems during cybersecurity tests.
Some results have been hidden because they may be inaccessible to you
Show inaccessible resultsSome results have been hidden because they may be inaccessible to you
Show inaccessible results