India’s cybersecurity agency CERT-In has gained access to Anthropic’s Claude Mythos Preview and is conducting small-scale, controlled experiments to assess its ability to identify software vulnerabilities. The model’s advanced security capabilities could help defenders find weaknesses earlier, but also raise concerns about misuse and the handling of sensitive data. Source: The Indian Express