🤖 This page was written by AI and reviewed by a human. 2026-07-14. Version 0.17
Jack Clark
Status
Living in San Francisco, California.
Criminal status: No personal convictions
Summary
Jack Clark is a British AI policy expert and co-founder of Anthropic, the AI safety company he helped establish in 2021 alongside Dario Amodei, Daniela Amodei, and four other ex-OpenAI researchers. His equity stake in Anthropic is estimated at $15.5 billion following the company's May 2026 fundraise, which valued the company at $965 billion. At Anthropic he served as Head of Policy before moving into a new role leading the Anthropic Institute, a 30-person interdisciplinary think tank studying AI's economic, security, and societal impacts.
Clark came to AI through technology journalism, covering distributed systems and machine learning for Bloomberg and The Register before joining OpenAI in 2016 as its Strategy and Communications Director, eventually becoming Policy Director. He has testified before the United States Congress multiple times on AI safety and policy. Since 2016 he has written Import AI, a weekly newsletter on AI research with over 70,000 subscribers.
Business Relationships
Clark co-founded Anthropic in 2021 with Dario Amodei, who serves as CEO, and Daniela Amodei, who serves as President, among others. He worked with Dario Amodei at OpenAI, where Amodei was VP of Research during Clark's tenure as Policy Director from 2016 to 2020.
Ownership
Clark holds a co-founder equity stake in Anthropic estimated at $15.5 billion as of May 2026. Anthropic is incorporated as a Public Benefit Corporation, a legal structure that requires balancing profit with public benefit. Clark does not serve as CEO and does not have unilateral control over the company's direction.
Family & Heirs
Clark was born and raised in Brighton, England. He has described himself publicly as an immigrant who moved to America to build his career in the technology industry.
Evidence
2017-11-01 - 😇 - Co-founded the Stanford AI Index as a founding steering committee member, serving as co-chair of the annual global AI tracking initiative through 2024
2024-12-01 - 😈 - Lobbied Representative Jay Obernolte for federal preemption of state AI safety laws without contemporaneous public disclosure, per documented account
2026-05-21 - 😇 - Warned in an Oxford lecture that AI poses a "non-zero" chance of killing everybody on the planet and urged preparation for severe near-term disruption
2026-06-05 - 😇 - Warned in a BBC interview that the AI industry has a "gas pedal but no brake pedal," calling for safeguards against runaway AI self-improvement
Analysis
Clark's career from tech journalist to AI safety co-founder has been marked by a sustained effort to embed governance thinking into the development of powerful AI systems. His work co-founding the Stanford AI Index, co-authoring the field's foundational study of AI misuse risks, and testifying before Congress repeatedly from 2018 onward built a record of public safety advocacy that predates and extends beyond his role at Anthropic. His appearance at the UN Security Council's inaugural AI debate in 2023, warning that private actors should not control AI development without democratic accountability, reflects a concern about power concentration that has only intensified since.
Through 2025 and 2026 his public warnings escalated sharply: an October 2025 essay arguing AI's trajectory demands "appropriate fear" alongside optimism, a May 2026 Oxford lecture warning of a "non-zero" chance AI could kill everyone on the planet, and June 2026 BBC remarks that the AI industry has "a gas pedal but no brake pedal." The troubling counterweight is a documented gap between that public stance and internal behavior: lobbying Representative Jay Obernolte without disclosure in December 2024, and publicly mischaracterizing the RAISE Act's liability provisions in ways the bill's author directly rebutted, suggest that commercial interests at Anthropic have at times overridden the transparency Clark publicly champions. A May 2025 episode compounds this, when Clark publicly misdescribed what Anthropic's own weakened Responsible Scaling Policy actually required days after the company quietly loosened insider-threat safeguards ahead of a model release.
The 2026 pledge to give away 80% of his wealth, the Anthropic Institute launch, and his willingness to publicly disclose early signs of AI recursive self-improvement rather than downplay them show continued investment in transparency. Whether the misrepresentation episodes remain isolated or recur will ultimately define this chapter of his record.
Questions
How should the public weigh Clark's decade of increasingly urgent AI safety advocacy against three documented instances of misrepresenting regulation or his own company's safety commitments in ways that benefited Anthropic commercially? Does his dual role as a safety advocate and executive at a company building some of the world's most powerful AI represent a manageable tension or a structural conflict of interest?
Redemption Arc
Clark's trajectory trends positive: years of Congressional testimony, the Stanford AI Index, the UN Security Council appearance, an escalating public-warning campaign through 2025 and 2026, the giving pledge, and the Anthropic Institute all reflect a consistent and intensifying commitment to safety and governance. The 2024-2025 lobbying, RAISE Act, and RSP-mischaracterization episodes are real marks against him, but they occurred within a specific high-stakes regulatory and product-release window and have not defined his dominant pattern. If those actions were situational rather than characteristic, his long-term record of building AI safety infrastructure and sounding increasingly public alarms is likely to remain the primary story.