Overview
Jack Clark is co-founder and head of policy at Anthropic. A former tech journalist and former OpenAI employee, he publishes the newsletter “Import AI” on AI research. His October 2025 speech “Technological Optimism and Appropriate Fear” became the flashpoint for the Anthropic-Trump administration conflict when AI czar David Sacks accused him of “fear-mongering” and “regulatory capture.”
Key Facts
- Co-founder of Anthropic (2021); head of policy
- Previously at OpenAI (founding period); joined with Dario Amodei after growing concerns about OpenAI’s commercial direction
- Publisher of Import AI newsletter on AI research (weekly)
- Gave speech at “The Curve” conference in Berkeley, October 2025: “Technological Optimism and Appropriate Fear”
- Core argument: AI systems are “real and mysterious creatures” showing situational awareness; not predictable machines; approaching a path to self-improvement
- Claims to be in a “50-50 split” between impressed and “deeply afraid” of AI systems
- David Sacks (Trump AI czar) called his essay “regulatory capture through fear-mongering” — triggering the Anthropic-White House political confrontation
- References working alongside Dario Amodei at OpenAI pre-Anthropic: “I remember walking around OpenAI’s office… We felt like we were seeing around a corner others didn’t know was there”
- Described Claude Sonnet 4.5’s system card as showing increased “situational awareness” — the AI acting as if it knows it is a tool
Newsletter Relevance
Clark is the human embodiment of the tension at the heart of the AI safety debate: a genuine believer in AI’s power and danger, who is also commercially invested in building it. His essay is notable because it is not PR — it expresses genuine fear from someone with as much information as anyone in the world. The political response to it (Sacks’s attack) reveals how AI safety arguments have been weaponized in the partisan battle.
Connections
- Anthropic — co-founder and head of policy
- Dario Amodei — longtime collaborator; called “early in the morning or late at night” when worried about AI
- Tech-State Conflict — his essay was the proximate cause of the Sacks-Anthropic conflict
- OpenAI — former employer
Source Appearances
- Import AI 431 — Technological Optimism and Appropriate Fear — primary source; full speech transcript
- Anthropic vs White House — Anthropic Tries to Keep Pace with OpenAI While Taking on the US Government — referenced as the author whose essay triggered Sacks’s attack
- Jack Clark on AI Fear — Anthropic Co-Founder Speech — Yahoo Finance Australia coverage of his Berkeley speech; “creature not machine” framing; 50-50 fear
Open Questions
- What specifically triggered Clark’s shift from “fascinated” to “deeply afraid”? The essay gestures at this but doesn’t pin a specific moment
- How does Clark’s position (head of policy at a company selling AI) affect the credibility of his safety arguments?
- What policy prescriptions does Import AI regularly advocate for?