Leaked documents reveal Anthropic’s next AI model, Mythos, with major gains in coding and cybersecurity, raising new questions about AI risks.
Two critical-severity ShareFile vulnerabilities allow remote attackers to bypass authentication and execute arbitrary code.