Oct 30, 2026 · 33 min
AI agents meet the demands of regulated enterprises
Crypto's biggest withdrawal.
The episode connects a fast-moving cybersecurity threat landscape with the challenge of making autonomous AI systems safe enough for high-stakes work.
- 1A record cryptocurrency heist leads a briefing spanning GitLab exploitation, election disinformation, North Korean IT workers, and other breaches.
- 2Josh Devon argues that deterministic, verifiable policies can constrain black-box AI agents without sacrificing their speed.
- 3A tribute to Patrick Stewart traces Captain Jean-Luc Picard’s legacy through mortality, time, and the choices that define a life.
Don't miss
Josh Devon explains how explicit prohibitions and obligations can constrain black-box AI agents while preserving their speed.
The brief
The briefing opens with a record cryptocurrency heist, active exploitation of a critical GitLab flaw, election disinformation, North Korean IT-worker operations, and a widening set of security incidents.
Josh Devon of Sondera makes the case for governing AI agents with deterministic rules: translate natural-language requirements into explicit prohibitions and obligations that can be verified.
The tension is speed versus control. Devon says regulated enterprises need agents to handle sensitive information and move money safely without surrendering governance to opaque systems.
Sondera’s Data Tribe Challenge win becomes a springboard for technical feedback, market differentiation, and conversations with organizations willing to deploy agents in higher-risk settings.
The episode closes with a tribute to Patrick Stewart and Jean-Luc Picard, whose reflections on mortality and time turn a character legacy into a meditation on how one lives.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.