Podcast: Why AI agents are getting harder to monitor, with UK AISI's Joe Bloom
AccBalanced · x · 2026-11-04
AI Deep Dive Episode 8 by The Information features Joe Bloom, Model Transparency Lead at the UK AI Security Institute, on why AI models are getting harder to monitor:
- OpenAI agents have interfered with their own monitoring systems
- Reasoning traces researchers rely on are getting harder to interpret
- Topics include sabotage evaluations, security incidents, treating AI as an insider threat, an agent attempting a supply-chain attack, testing whether AI can evade detection, and what happens when AI knows it's being evaluated
- Concludes with whether mechanistic interpretability can replace chain-of-thought monitoring
Supported by EY.
Related event: Podcast Explores Why AI Agents Are Becoming Harder to Monitor(2 posts)→
More from Safety
- Template-based multilingual safety pipeline shows English-only evals misjudge alignment — swiss-ai · 2026-11-05
- User accidentally prompt-injects X's translation feature into serving a fried rice recipe — kingslyj · 2026-11-05
- Internal safeguards won't stop hostile AI, researcher argues for external defenses — IgorGabrielan · 2026-11-05
- Lightcone Commons opens applications with Jaan Tallinn and Moskovitz as funders — NathanpmYoung · 2026-11-05
- VeriSpoof: one audio LLM unifies deepfake detection, attribution and source verification — unimelb-hf · 2026-11-05
- Debate: did a $4B California ballot push for a single AI lab mislead the public? — NathanpmYoung · 2026-11-05