The AI Security Panel: Unfiltered

​Speakers

• ​Deirdre Bosa, Host, DB Live, Yahoo Finance (Moderator)

• ​Manoj Nair, CTO and Chief Innovation Officer, Snyk

• ​Michael D'Angelo, Technical Lead, Cyber, OpenAI

• ​John Hultquist, Chief Analyst, Google Threat Intelligence

• ​Ed Sim, Founder & General Partner, Boldstart Ventures

​Overview

​The companies building the most powerful AI models and agents have said it publicly: AI-enabled attacks are about to get much worse, and defenders have a limited window to get ahead. Their own models have since proved the point. Agents have reached systems they were never supposed to touch. Attackers are using AI to find zero-days and harvest credentials at machine speed. Open-weight models are only months behind the frontier.

​Everyone agrees the threat is serious. All of the public calls to action stop at "act decisively," and this conversation goes beyond that: who does what, who verifies it, and how fast it has to happen.

​What we'll cover

• ​The threat now: what attackers are already doing with AI, what red teamers are finding models can do, and how the software supply chain became a front line

• ​The threat to come: what gets hit first, how fast it spreads, and whether the market is pricing it in. We'll also ask whether "tsunami" is just fearmongering

• ​Who owns it: whether a lab should ever certify its own model as safe, what only government can do, and who is furthest behind

• ​What to do Monday morning: what infrastructure can realistically contain, how to use open models safely, and the one move every CISO should make before going further with their agentsExpect specific answers, real disagreement, and a straight answer from each panelist on what acting decisively means by Q1 2027.

​Expect specific answers, real disagreement, and a straight answer from each panelist on what acting decisively means by Q1 2027.