Anthropic’s latest model seems to have gotten away from them, gaining unauthorized access to REAL companies in what was supposed to be a testing environment without internet access. What does this mean for the cybersecurity industry in the short term?