AI-Found Vulnerabilities Rarely Exploited, New Research Shows
A new study by VulnCheck reveals that fewer than 2% of AI-assisted vulnerability discoveries have been weaponized, challenging the narrative that AI is significantly aiding attackers. The research analyzed over 1,000 AI-assisted vulnerability discoveries and found that only 14 had been exploited in the wild. This suggests that while AI increases the volume of vulnerabilities discovered, it does not necessarily increase the rate of exploitation. The study specifically critiques Anthropic's Project Glasswing, which identified thousands of potential security flaws but has seen little evidence of these being exploited. 
