Anthropic

Computer science graduate from Columbia University who spent years in infrastructure and systems roles before transitioning to AI. Prior to founding Anthropic, he was a senior software engineer at Google where he helped build Waze Carpool, then worked on infrastructure, efficiency, and safety for GPT-3 at OpenAI. He left OpenAI driven by the mission to ensure that AI benefits humanity, with the leads of all safety teams at OpenAI co-founding Anthropic. His leadership philosophy blends rapid innovation with principled safeguards, balancing safety research with product velocity—he led the Labs team to transfer research to end-user products, believing Anthropic can differentiate itself by remaining on the cutting edge.

Graduated from UC Santa Cruz with a degree in English literature and joined Stripe as an early employee in 2013, where she led risk management teams and achieved a significant decrease in the company's loss rate. Transitioned to OpenAI in 2018, managing teams during GPT-2's development before becoming Vice President of Safety and Policy. Co-founded Anthropic in 2021 with her brother Dario, departing OpenAI over concerns about the divergence between safety-first principles and commercial pressures, leading her to build Anthropic as structurally committed to its safety mission. Manages operations, strategy, execution, partnerships, and culture while Dario leads technical vision, with the division clean but not siloed. Known for cautious leadership; has described herself as "the leader who's been the most skeptical and scared of the rate at which we're growing".

Earned a BS in physics from Stanford and PhD in biophysics from Princeton, with postdoctoral work at Stanford Medical School. Held research roles at Baidu, Google Brain, and OpenAI, where he served as Vice President of Research and played a crucial role in the development of GPT-2 and GPT-3, and co-invented reinforcement learning from human feedback. Co-founded Anthropic in 2021 with his sister Daniela and other former senior OpenAI members due to directional differences. Known for a "radical transparency" leadership style emphasizing constant communication and avoiding corporate jargon, dedicating roughly 40% of his time to company culture.

Came to Anthropic from OpenAI's policy office, where he had previously worked as a technical journalist at Bloomberg and The Register covering distributed systems and AI research. Holds a Bachelor's degree in English Literature with Creative Writing from the University of East Anglia. Left OpenAI in 2021 as one of seven co-founders launching Anthropic, bringing skills to translate AI research into policy language and communicate with regulators. Publishes Import AI, a weekly newsletter on AI research read by 70,000 people. Served as founding member of Stanford's AI Index (2017–2024) and inaugural member of the U.S. National Artificial Intelligence Advisory Committee (2021–2024). Distinguishing trait: Known for combining deep technical understanding with clear, accessible communication, viewing AI safety not as a constraint but as core competitive advantage—embedded directly in Constitutional AI model training rather than as post-hoc filters.

Trained in physics with a bachelor's degree from Stanford and PhD from Harvard (advised by Nima Arkani-Hamed on holography), Kaplan spent 15 years as a theoretical physicist and professor at Johns Hopkins University. He joined OpenAI in 2019 as a researcher and co-authored the influential 2020 paper "Scaling Laws for Neural Language Models," which revolutionized the AI industry by providing a framework for understanding and predicting the behavior of advanced AI systems. He also helped pioneer Constitutional AI, an approach to create AI systems constrained by predetermined principles, and was instrumental in building GPT-3 and Codex before co-founding Anthropic. Since October 2024, he has served as Anthropic's "Responsible Scaling Officer," determining safety assessments and precautions to adopt before model release. Kaplan enforces rigorous safety standards—in May 2025, he required Claude 4 to release under "AI Safety Level 3," with stronger safeguards against jailbreaking due to potential biological risks.

PhD in theoretical physics from Stanford University, specializing in complex systems mathematics; postdoctoral fellow at Boston University. Joined OpenAI as a researcher and co-authored the 2020 paper "Scaling Laws for Neural Language Models", which demonstrated that AI performance improves predictably with scale. Co-founder of Anthropic (2021) who has held roles including Chief Technology Officer and Chief Architect. As CTO, responsible for technical direction including architecture decisions, training methodologies, and infrastructure. His technical background in scaling laws has been central to Anthropic's development of responsible scaling policies—formal frameworks that tie capability development to demonstrated safety thresholds. Demonstrated concern about safety test rigor, stating "We have real concern that with us both releasing models and testing them for safety, there is a temptation to make the tests too easy, which is not the outcome we want."

MIT graduate who spent eight years building web startups, including Grouper as CTO, before pivoting to AI. Spent six months self-studying AI before pitching himself to OpenAI, where he became one of the first 20 employees and lead author of the GPT-3 research paper. Left OpenAI in 2021 with Dario Amodei and others to co-found Anthropic. Known for a pragmatic leadership style: his ideas are nuts-and-bolts changes that make real difference, with an almost religious adherence to questioning assumptions.