🤖 This page was written by AI and reviewed by a human. 2026-07-14. Version 0.17
Jared Kaplan
Status
Born in 1983. Living in Pacifica, California.
Criminal status: No personal convictions
Summary
Jared Kaplan is a theoretical physicist and AI researcher who co-founded Anthropic in 2021 alongside six other former OpenAI employees, explicitly over concerns about safety at frontier AI labs. He serves as the company's chief science officer and, since October 2024, as its Responsible Scaling Officer — the executive responsible for determining what safety measures must be in place before any new model is released.
Before founding Anthropic, Kaplan spent a decade as a physics professor at Johns Hopkins University, focusing on quantum gravity and conformal field theory. At OpenAI from 2019 to 2021, he co-authored the Scaling Laws for Neural Language Models paper and the GPT-3 paper, two of the most cited works in modern AI. At Anthropic, he helped develop Constitutional AI, a training framework that uses a written set of principles to guide model behavior rather than relying solely on human feedback labels.
Forbes estimated each Anthropic co-founder's net worth at $15.5 billion in May 2026 after the company raised $65 billion at a valuation of approximately $965 billion, making Anthropic the most valuable AI startup in the world.
Business Relationships
Kaplan co-founded Anthropic with Dario Amodei, Daniela Amodei, Jack Clark, Chris Olah, Tom Brown, and Sam McCandlish. He and Dario Amodei previously worked together at OpenAI, where their research group developed scaling laws and co-authored the GPT-3 paper.
Ownership
Kaplan holds a founding equity stake in Anthropic, estimated at just over 1.6% of the company as of May 2026. Anthropic has major investment partnerships with Amazon and Google, filed a confidential S-1 registration statement with the SEC on June 1, 2026, and has selected investment banks for a potential IPO. He is listed as currently on leave from his Johns Hopkins associate professor position.
Family & Heirs
Kaplan lives in Pacifica, California, and has a son.
Evidence
2014-03-06 - 😇 - Awarded 2014 Sloan Research Fellowship for particle physics research at Johns Hopkins
2015-07-01 - 😇 - Awarded NSF CAREER grant for Conformal Field Theory and Quantum Gravity research
2020-01-23 - 😇 - Co-authored Scaling Laws for Neural Language Models, foundational AI science research
2020-05-28 - 😇 - Co-authored GPT-3: Language Models are Few-Shot Learners
2022-12-15 - 😇 - Submitted Constitutional AI: Harmlessness from AI Feedback, pioneering AI alignment research
2023-12-06 - 😇 - Testified to US Senate on AI risk, alignment, and guarding against catastrophic scenarios
2024-10-15 - 😇 - Named Anthropic Responsible Scaling Officer with authority over safety decisions before model release
2025-02-24 - 😇 - Co-authored Forecasting Rare Language Model Behaviors, AI safety research predicting dangerous outputs at deployment scale
2025-05-08 - 😇 - Co-authored Reasoning Models Don't Always Say What They Think, AI safety research on chain-of-thought faithfulness
2025-05-22 - 😇 - Enforced AI Safety Level 3 protections on Claude Opus 4 over bioweapons risk
2025-08-26 - 😇 - Named to TIME 100 Most Influential People in AI 2025 for enforcing safety standards at frontier AI lab
2025-09-05 - 😈 - Anthropic, co-founded by Kaplan, agreed to pay $1.5B to settle lawsuit over training Claude on 7 million pirated books
2025-12-13 - 😇 - Publicly warned humanity faces decision on whether to allow AI recursive self-improvement by 2027-2030
2026-01-27 - 😇 - Pledged 80% of personal wealth to combat AI-driven inequality alongside all Anthropic co-founders
2026-02-24 - 😈 - As Responsible Scaling Officer, authorized dropping hard safety limits from Anthropic RSP, citing competition
Analysis
Kaplan's record is dominated by a sustained, decades-long commitment to science and AI safety research. His academic contributions — from theoretical physics at Johns Hopkins to the scaling laws paper and Constitutional AI framework — have directly shaped how the field understands and builds advanced AI systems. His continued research output through 2025, including work on predicting dangerous AI behaviors at deployment scale and evaluating whether AI reasoning is faithful to its actual processing, reflects an ongoing scientific engagement with AI safety problems rather than a departure from research into purely executive work.
His active enforcement of AI Safety Level 3 measures on Claude Opus 4 in May 2025 — for which TIME named him to its 100 Most Influential People in AI list — and his Senate testimony advocating for safety regulation demonstrate that his safety work extends beyond research papers into institutional and policy action. In April 2026, he personally announced Project Glasswing, a coalition with Amazon, Apple, Google, and Microsoft that used an unreleased Anthropic model to find and disclose critical software vulnerabilities before they could be weaponized.
The significant negative is his decision in February 2026, as Responsible Scaling Officer, to drop the hard safety limits that had been Anthropic's defining public commitment since 2023 — arguing that unilateral pauses while competitors continue building would not help anyone. That justification is contested: critics argued the change represented a capitulation to competitive pressure. The $1.5 billion settlement over pirated training data also reflects conduct within the company Kaplan co-founded and leads scientifically.
Questions
The RSP rollback raises a fundamental question: when an AI lab drops its most visible safety commitment citing competitor behavior, has safety become a differentiating principle or a marketing claim? Kaplan's framing — that continuing to build is safer than pausing unilaterally — is internally coherent but also convenient for a company with a significant founder stake in continued model development. The 80% wealth pledge is now worth over $12 billion at Kaplan's current valuation: will it translate into actual disbursements that can be evaluated, or remain a forward commitment?
Redemption Arc
Kaplan's trajectory is difficult to read. In the same year he enforced stricter safety standards on Claude Opus 4, published safety research on predicting rare dangerous behaviors, and publicly warned about AI's existential risks, he also dismantled the hard safety stop that had been Anthropic's core public promise. Both moves are consistent with a view that safety requires expert judgment rather than bright-line rules — but they are also consistent with a pattern of adjusting standards under competitive pressure while maintaining a safety-focused public identity.