Why AI Agent Access Is Security’s Big Unsolved Problem ft. Sean Cassidy, CISO at Plaid

Sean Cassidy (Chief Information Security Officer @ Plaid) says the Hugging Face attack split security into a before and after. Security budgets used to track how valuable you were to a motivated attacker. Now an agent can hack you for a few tokens, sometimes by accident. On this episode of Early Adopters, Sean breaks down: Why AI agent identity and access is still unsolved How agents make a central SIEM less important How to make risk acceptance a conscious leadership decision How Plaid decides what security tooling to buy vs. build What founders get wrong when pitching CISOs CHAPTERS 00:00 Intro 03:50 Building A Cloud SIEM For Banks 09:17 The LostPass Phishing Attack 11:50 Scaling Asana Security For Enterprise Buyers 16:42 Making Risk Acceptance A Conscious Decision 21:25 Why Agents Make SIEMs Less Important 22:55 The Hugging Face Attack And Agents As Attackers 27:54 Founder Pitches And Buy vs. Build Decisions 33:19 Why Agent Access And Identity Is Unsolved