AI Models Discover Zero-Day Vulnerabilities at Unprecedented Rates

Recent advancements in AI, particularly with models like Anthropic’s Claude Opus 4.6, have led to the discovery of over 500 high-severity vulnerabilities in open source codebases. These developments indicate a shift in cybersecurity, where AI is now capable of identifying critical zero-day vulnerabilities that were previously the domain of elite human hackers. Organizations must now focus on exposure prioritization and AI-driven remediation to effectively manage these vulnerabilities.