Anthropic’s cybersecurity-focused AI model is reportedly helping U.S. officials identify software vulnerabilities across government systems, marking a notable shift after the company weathered a contentious dispute with the Donald Trump administration.