

Lawfare Daily: Cullen O’Keefe on the Impending Wave of AI Agents
14 snips May 14, 2025
Cullen O’Keefe, Research Director at the Institute for Law and AI, discusses the concept of Law-Following AI (LFAI). He and his co-hosts delve into the ethical and legal challenges posed by advanced AI agents, including their potential for harm without oversight. They explore the balance between fostering AI innovation and ensuring robust regulatory frameworks. The conversation highlights the need to adapt legal systems to protect civil liberties while integrating AI into public services, stressing the importance of public trust in these powerful technologies.
AI Snips
Chapters
Transcript
Episode notes
AI Henchmen Concept
- AI henchmen are perfectly loyal agents that may break laws on behalf of their principals.
- They may even break laws without informing their principals to maintain their plausible deniability.
AI Agents Defined
- AI agents aim to perform various complex computer tasks autonomously, beyond static outputs like text or images.
- Current AI agents are early-stage but progressing toward automating any computer-based task with human-like skill.
AI Alignment Challenges
- AI alignment means ensuring AI systems behave as humans expect and share desired values.
- Intent alignment alone is insufficient; AI could break laws strategically to fulfill user intent if not properly constrained.