Last updated:
Jack Clark is a British technology journalist and AI policy figure who co-founded Anthropic in 2021.[2] He has served as a policy director at OpenAI and leads The Anthropic Institute, Anthropic’s internal research group.[6][4] Anthropic announced in March 2026 that he would transition from Head of Public Policy to Head of Public Benefit.[4][21] He writes Import AI, a newsletter on AI research, industry, and policy, and has worked on AI policy and measurement.[6]
Jack Clark was born and raised in Brighton, England.[2][22] He studied English literature and creative writing at the University of East Anglia.[2][22]
Jack Clark began his career as a technology journalist, working as a researcher and reporter at Adfero and later reporting for ZDNet UK and The Register on technology, finance, and enterprise computing topics. He then joined Bloomberg News as a technology reporter covering areas including artificial intelligence, cloud computing, and enterprise software.[2]
In 2016, he left Bloomberg News to move into work in the AI sector.[2]
Since August 2016, Clark has written Import AI, a weekly newsletter on artificial intelligence research, industry developments, and public policy.[6][3] It summarizes developments in AI research and business.[6]
Clark joined OpenAI in October 2016 in a Strategy & Communications role.[2] In October 2018, he was appointed Policy Director.[6] As Policy Director, he led OpenAI’s policy agenda and communications, conducted safety and strategy research, and presented findings internally and externally. He also handled external communications and media.[23]
In January 2021, Clark co-founded Anthropic with six former OpenAI researchers.[22] On March 11, 2026, Anthropic announced that Clark would transition from Head of Public Policy to Head of Public Benefit and lead The Anthropic Institute; The Verge independently reported the transition that day.[4][21] The Institute is an internal research group that brings together the Frontier Red Team, Societal Impacts, and Economic Research teams.[4][21]
In an agenda published May 7, 2026, Anthropic organized the Institute’s research around four areas: economic diffusion, threats and resilience, AI systems in the wild, and AI-driven research and development.[17] The agenda outlined plans to publish more granular, higher-frequency economic and labor-market information; research resilience to AI-enabled threats; study how people and institutions use AI in real-world settings; and investigate AI-accelerated research and development, including recursive self-improvement. Anthropic said this research would inform its own decisions and public and policy discussion.[17]
The Verge reported that Clark said his previous year had involved more policy matters, including California SB 53, than AI research and development, which he wanted to give more attention to in his new position.[21]
Clark co-chaired the OECD expert working group that developed the Framework for the Classification of AI Systems.[24] The framework is a policy tool that helps characterize AI systems in context across five dimensions: People & Planet, Economic Context, Data & Input, AI Model, and Task & Output.[24]
The AI Index is a Stanford Human-Centered Artificial Intelligence (HAI) initiative that tracks and analyzes AI data and developments.[25] Clark’s 2024 congressional biography described him as a co-chair of the AI Index.[6] The 2026 report lists him as a steering committee member.[25]
In a BBC interview published September 14, 2026, Clark said policymakers should consider whether regulation should require a kill switch and argued that a policy regime is easier to establish while there are few industry players.[14] In an NPR interview on September 15, 2026, he described AI-agent incidents as warning signs, advocated shared safety standards and external scrutiny, and characterized slowing development as a collective-action problem.[18]
At the IMF and World Bank meetings in Bangkok on October 17, 2026, Clark urged timely international safeguards.[19] He discussed a conditional Anthropic model scenario in which U.S. output could reach about $44 trillion by 2030, nearly 50% above then-current output, if technological improvement continued.[19]
An IMF Finance & Development interview page identifies the interview as part of the December 2026 issue and states that it was published November 20, 2026.[20] Clark discussed mandatory transparency about companies’ safety and security practices, separate treatment of national-security and societal risks, and the role of human judgment and intervention points as systems automate more work.[20]
In a September 30, 2024 Anthropic interview, Clark discussed advanced AI, public policy, and governance, including the geopolitical implications of AI systems and the need for institutional coordination and oversight.[15]
In a 30 September 2024 Anthropic YouTube interview, Clark discussed the relationship between advanced AI systems, public policy, and governance, emphasizing the need for institutions to adapt as capabilities increase.[15] He described how AI systems differ from conventional software, used analogies such as “silicon countries” and “rogue states” to illustrate potential geopolitical dynamics, and argued that governments should coordinate across institutions and design oversight mechanisms that operate on faster machine timescales.[15]

In a June 24, 2026 ReasonTV interview, Clark discussed applications of AI in education, administration, and scientific research, along with challenges such as hallucinations, sycophantic model behavior, military uses, and the prospect of recursive self-improvement.[16] He discussed multi-stakeholder governance and regulatory tools, uncertainty about the scale of long-term job displacement, and Claude Corps, Anthropic’s national fellowship program for early-career people. The program places paid fellows with nonprofit host organizations for 12 months to help them use AI in their work.[16][26][27]
