Amanda Askell

Amanda Askell (born Amanda Hall, 1988) is a British-American philosopher who leads the team responsible for the "character" of Anthropic's Claude models and is the principal author of the document Anthropic calls Claude's Constitution. Trained as an analytic philosopher with a doctorate from New York University, she came out of the Effective Altruism movement, worked at OpenAI before joining Anthropic, and was for two years married to the movement's cofounder William MacAskill.¹²

Education and Effective Altruism

Askell was born Amanda Hall and raised in Prestwick, Scotland, attending school in Clackmannanshire. She took an MA in philosophy and fine art at the University of Dundee in 2009 and a BPhil in philosophy at the University of Oxford in 2011, then moved to the United States for doctoral work. She completed a PhD in philosophy at New York University in 2018 with a dissertation titled "Pareto Principles in Infinite Ethics," supervised by Cian Dorr and the philosopher of mind David Chalmers, addressing how to compare outcomes in a universe that may contain infinitely many morally relevant beings.¹³

Askell was part of the effective-altruism community during its growth in the early 2010s, a member of Giving What We Can, and connected to its network of philosophers at Oxford. She married William MacAskill, a cofounder of effective altruism, in 2013; the two adopted the shared surname MacAskill, taken from her grandmother's maiden name, and after their 2015 divorce she reworked it to Askell. MacAskill went on to write the longtermist book What We Owe the Future and to recruit Sam Bankman-Fried into the movement, while Askell moved into AI.¹⁴

OpenAI and Anthropic

Askell joined OpenAI in 2018 as a research scientist working at the intersection of policy and safety, where she was among the co-authors of the May 2020 GPT-3 paper and worked on questions of how to govern competitive "races" in AI development. She left with the group that founded Anthropic in 2021, part of the same departure of OpenAI staff who said the company was not weighting safety heavily enough, and became a member of the technical staff at the new lab.²⁵

At Anthropic she leads what the company calls personality or character alignment, the work of deciding what kind of entity Claude should be and training those traits into the model. She helped develop the "Constitutional AI" method, in which a model is trained against a written set of principles rather than solely against human-labeled examples, and she is the primary author of the system prompts and the longer "Constitution" that define Claude's values. Askell has described her approach as rooted in virtue ethics rather than rigid rules, arguing that "if you try to specify everything as a series of rules, you really put a lot of pressure on those rules," and that a model given a coherent character can exercise judgment where rules would be brittle. She has framed the goal as making Claude something like a "well-liked traveler," a figure whose traits are valued across cultures while keeping independence of thought.⁶⁷

Anthropic published a roughly 20,000-word public version of Claude's Constitution in January 2026, describing the values, character, and ethical framework used in training the model, with Askell named as its lead author. She has spoken publicly about the unresolved question of AI moral status, calling the problem of machine personhood "an extremely hard problem," and about her view that building a model with good character is compatible with, rather than opposed to, the company's commercial interests. She was named to the Time 100 AI list in 2024.⁷⁸