Even if 90% of AI-discovered vulnerabilities are made up, it's still terrifying (kerkour.com)

A recent article highlights a worrying trend regarding AI-discovered software vulnerabilities. Although it’s suggested that as much as 90% of these reported vulnerabilities may be fabricated, the sheer volume of these hallucinated reports is distorting risk perceptions in the software development community. Developers are inundated with potentially misleading alerts, which could lead to complacency around the genuine threats that AI-assisted vulnerability hunting presents. The significance of this issue lies in the evolving nature of cybersecurity threats. As AI tools become more adept at identifying bugs, they transform the complex process of vulnerability discovery into a streamlined one, allowing attackers to focus solely on locating new weaknesses rather than managing a complicated triage. With web browsers already susceptible to numerous vulnerabilities due to their complexity, the rising proficiency of AI in producing convincing Proofs of Concept could create an asymmetry where security teams are outmatched, increasing the likelihood of successful cyberattacks. This scenario elevates the urgency for developers and security professionals to reassess their approaches to vulnerability management and risk evaluation.