🎙️SECURITY.COM The Podcast: Known Exploited Doesn’t Stay Exploited
CISA KEVs limitations, deterministic ML, and the case for localized vulnerability predictions
Empirical Security founders Ed Bellis, Jay Jacobs, and Michael Roytman join SECURITY.COM The Podcast to discuss what happens when vulnerability prioritization relies on signals that don’t keep up with attacker behavior.
Here are the highlights
- Why “known exploited” isn’t a permanent state
- What makes CISA’s KEV the floor, not the finish line
- Why LLMs can’t do the job of a predictive model
- How positive unlabeled (PU) modeling helps quantify the unknown
- Why local models need to answer “yeah, but why?”
For a closer look at what AI-accelerated vulnerability management means for defenders, read Every Defender Deserves Frontier AI.
Find it wherever you listen
✔️ YouTube
✔️ Spotify
✔️ Apple Podcast
