Future of Life Institute Podcast Defense in Depth: Layered Strategies Against AI Risk (with Li-Lian Ang)
14 snips
Apr 2, 2026 Li-Lian Ang, a Blue Dot Impact team member building a workforce to reduce AI risks. She discusses defense-in-depth strategies in three layers. Topics include AI-enabled bio threats, automated cyberattacks by agents, economic disempowerment and power concentration, and how society can detect, prevent, and withstand harms.
AI Snips
Chapters
Transcript
Episode notes
AI Lowers Barriers For Bio And Cyber Threats
- AI lowers barriers for engineered pandemics and large-scale cyberattacks by giving novices biological and offensive cyber knowledge.
- Ang cites AI-assisted molecule design and Anthropic’s report on AI agents used in a cyber attack as concrete examples.
Agents Let One Operator Run End To End Cyberattacks
- AI agents can replace multiple human roles in cyberattacks by automating vulnerability discovery, exploitation, payload delivery, and coordination.
- Ang emphasizes one person plus a powerful agent can run end-to-end attacks, changing attacker-resource calculus.
Defence Becomes Agent Arms Race Versus Usability Tradeoffs
- Defense will become an arms race between attacker and defender agents; effectiveness depends on whose agent is better and who has more resources.
- Ang notes this mirrors existing cyber dynamics but may require new formal verification or trade-offs in usability.