Anthropic's Project Glasswing Identifies Over 10,000 Vulnerabilities in One Month

📌 Anthropic’s AI-driven cybersecurity initiative, Project Glasswing, identified over 10,000 high- or critical-severity vulnerabilities within one month of its launch. The project, built around the Claude Mythos Preview model, was announced by Anthropic on Friday as part of its defensive security efforts. No specific CVEs, affected systems, or technical details of the vulnerabilities were disclosed in the report. The findings highlight a significant gap in patching practices, though no direct impacts or exploitation cases were mentioned. The program’s timeline is limited to the month following its activation, with no geographical or organizational scope specified.