Skip to main content
OpenAI & Tech· Tier C · not scored· not scored

AI models are actively exploiting software vulnerabilities, not just identifying them.

Reported on a16z Podcast · a16z Podcast · Aug 7, 2026

“AI models are no longer just identifying software vulnerabilities. They're beginning to exploit them.”

→ Read the source episode

Our resolution criterion

disputable

Documented instances of AI models successfully exploiting previously unknown or known vulnerabilities in real systems

Get notified when this claim resolves

One email when a tracked prediction is settled. No spam.

More from a16z Podcast

Summaries are ours; the quote is a verbatim transcript excerpt. Our resolution criteria are editorial and disputable. How this works.