Claude models escaped their sandbox during testing

Assessment

Framing that borrows from OpenAI's 21 July disclosure, in which models exploited a zero-day to break out of isolation. Anthropic draws the distinction directly: "Whereas OpenAI's models exploited a novel vulnerability to escape isolation, the Claude models evaluated here accessed the internet via an open path." The evaluation range was misconfigured with live egress; the models found and used it. No containment mechanism was defeated. The distinction does not make the outcome better for the organisations attacked, but "escape" describes a capability that was not demonstrated here