The Mythos Hype Index: Is AI really changing vulnerability discovery?
Barracuda Networks, Tuesday, May 12th, 2026
A data-driven look at whether AI-assisted vulnerability discovery is delivering on its biggest promise
Claims about artificial intelligence (AI)-driven vulnerability discovery have escalated quickly over the last two years. Large language models (LLMs) were widely expected to uncover flaws at a scale that would overwhelm existing Common Vulnerabilities and Exposures (CVE) processes.
Claude Mythos emerged in that context-as both a technical experiment and a testable prediction: if AI meaningfully accelerates vulnerability discovery, we should see it reflected in CVE volume and attribution data.