Snehal Antani is the CEO and co-founder of Horizon3.ai. Antani joins Big Technology to cut through the hype around models like Fable, Mythos, and GPT-5.6 in cybersecurity. Tune in to hear why AI agents fall for security decoys far more often than human hackers do, and why that flaw is actually a gift to defenders. We also cover how AI is speeding up exploit development from leaked patches, and why agents should be treated as insider threats. Hit play for a builder's-eye reality check on AI and cybersecurity. This is a 00:00 —AI and the new cybersecurity era 01:06 – Hype vs. reality on frontier model cyber capabilities 02:32 – AI agents click decoys 92% of the time 05:07 – What the decoys actually look like 07:20 – How AI speeds up reverse-engineering software patches 10:11 – Why unpatched vulnerabilities are the real danger window 13:23 – "Infinite cyber bullets" and the training data bottleneck 18:10 – Agents as "unlimited interns" for hackers 20:39 – Vibe-coded apps and the exploding attack surface 23:18 – Prompt injection as both attack and defense