Podcast: can AI data centers contain advanced AI? SL5's Lisa Thiergart on lab security
Miles_Brundage · x · 2026-11-10
The Information's AI Deep Dive Ep. 9 features Lisa Thiergart, co-founder and CEO of SL5 Task Force, on whether AI data centers can contain frontier models.
Topics: espionage and insider threats at AI labs, what Security Level 5 means, model theft and distillation, securing models pre-deployment, AI becoming the insider threat, neocloud security and rogue agents, and building a secure AI data center. Thiergart argues the industry should build logical controls by spring 2027, then physical controls.
More from Safety
- System card confirms Claude MyTHOS 5 and Fable 5 share the same underlying weights — repligate · 2026-11-10
- Median time from patch to first recorded exploitation was 51+ days every month of 2026, RootEvidence chart shows — WeldPond · 2026-11-10
- Anthropic's tightened limits frustrate independent researchers running open-science loops — omarsar0 · 2026-11-10
- Fireship breaks down how a restricted model found 948 root exploits in the Linux kernel — Fireship · 2026-11-10
- Sen. Ed Markey presses Google to let schools disable AI in Search — Polymarket · 2026-11-10
- No Priors: Securing Personal AI Agents in the Enterprise with Serval's Jake Stauch — No Priors · 2026-11-10