Who is Jack Clark?
Jack Clark is co-founder and head of policy at Anthropic, the AI safety company behind Claude. Before co-founding Anthropic, he served as policy director at OpenAI. He publishes Import AI, a widely-read weekly newsletter tracking developments across the AI industry, which has become an essential source for researchers, policymakers, and industry leaders.
Why He Matters for AI Work
Clark occupies a rare intersection: he’s deeply technical (co-founding one of the leading AI labs), deeply embedded in policy (shaping Anthropic’s government engagement and publishing research on AI monitoring), and deeply informed about the competitive landscape (through his newsletter). His perspective on AI agents is shaped by both building the technology and grappling with its societal implications.
He led the creation of the Anthropic Economic Index, one of the first systematic attempts to measure how AI is affecting different occupations and industries — giving economists outside AI companies data to work with.
Key Positions
- AI agents represent a fundamental shift from “talkers” to “doers” — systems that use tools and work autonomously over time
- Recursive self-improvement is happening at Anthropic (Claude Code building Claude) and requires extraordinary caution
- AI governance needs “systems of seeing” — monitoring infrastructure that makes AI’s impact visible to society
- Policy conversations require granular, state-level economic data to activate elected officials
- AI companies should not run ads, to avoid surveillance-capitalism dynamics with personal AI interactions
Related Reading
- AI Agents — The technology Clark is building and studying
- Recursive Self-Improvement — His top concern for 2026
- Knowledge Work Disruption — The economic shift he’s measuring
Video Mentions
An agent is something where you can give it some instruction and it goes away and does stuff for you. Kind of like working with a colleague.