Leo Zovic

Events

• Could Frontier Labs’ Internal Agents Already Go Rogue? 2026-06-04
• Testing LLM Cooperation in Multi-Agent Simulation 2026-04-16
• Adversarial Defenses for LLMs 2026-03-26
• The Anthropic-Pentagon Stand-Off 2026-03-24
• Open-weight LLMs – Their Strategic Import to Key AI Jurisdictions and The Growing Need for Safety Research, Dialogues, and Coordination 2026-03-10
• Network Topologies for AI and the Implications for Governance 2026-03-05
• AI Safety Thursday: Risks Emerging from Agent Swarms 2026-02-19
• AI Safety Thursday: Claude's New Constitution 2026-02-12
• AI Safety Thursday: Beyond Adversarial Robustness - Rethinking Sociopolitical Safety in AI Systems 2026-01-22
• AI Safety Thursdays: Why Attackers Are Winning and What We Can Do About It 2026-01-08
• AI Safety Thursday: Agentic Bug Detection - Progress and Deployment 2025-12-18
• AI Safety Thursday: Sandbagging - How Models Use Reward-Hacking to Downplay Their True Capabilities 2025-11-27
• AI Safety Thursday: Agentic property-based testing - finding bugs across the Python ecosystem 2025-11-13
• AI Safety Thursday: Monitoring LLMs for deceptive behaviour using probes 2025-11-06
• IABIED Reading Group: "Part 1: Nonhuman Minds" 2025-10-27
• AI Safety Thursday: The Limitations of Reinforcement Learning for LLMs in Achieving AI for Science 2025-10-23
• "If Anyone Builds It..." Reading Group 2025-10-20
• AI Safety Thursday: Modeling and Detecting Deceptive Alignment 2025-10-16
• AI Policy Tuesday: The Case for Regulating AI Companies, Not AI Models 2025-09-30
• AI Policy Tuesdays: Frontier AI Deployments in US National Security and Defence 2025-09-02