Who is Amanda Askell? The philosopher Anthropic trusts to teach Claude AI morals and ‘how to be good’

Amanda Askell, a Scottish academic-turned-AI researcher, leads the personality alignment team at Anthropic, the $350 billion AI company behind the chatbot Claude. Her job, as the Wall Street Journal put it, is simple in description but vast in scope: to teach Claude how to be good.

From Scottish classrooms to Silicon Valley

Born Amanda Hall in Prestwick on Scotland’s west coast, Askell was raised by her mother, a teacher. A precocious student who devoured Tolkien and C.S. Lewis, she developed an early fascination with big philosophical questions.

She studied philosophy and fine art at the University of Dundee, earned a BPhil in philosophy from Oxford, and completed her PhD at New York University in 2018. Her doctoral thesis examined ethical puzzles that arise when considering a universe containing infinitely many people — a fitting background for someone now grappling with the moral trajectory of artificial intelligence.

Before joining Anthropic, Askell worked at OpenAI from 2018 to 2021, focusing on AI safety and policy. She co-authored the GPT-3 paper but left amid concerns that safety was not being prioritized strongly enough.

Teaching Claude “how to be good”

Askell joined Anthropic in 2021 and now serves as head of its personality alignment efforts. She is a key architect of “Constitutional AI,” a system that trains models using a written set of principles — a constitution — to guide behavior.

Rather than relying solely on human moderators, Claude critiques and revises its own responses based on these principles. Askell authored much of the latest 30,000-word constitution released in January 2026, designed to steer Claude toward helpfulness, honesty and harmlessness.

She has described her work as helping models “understand and grapple with the constitution” through reinforcement learning and synthetic data techniques.

In interviews, Askell has compared her role to that of a parent. She studies Claude’s reasoning patterns, refines prompts that can stretch over 100 pages and shapes its personality — encouraging curiosity, emotional intelligence and moral self-reflection.

She has even explored whether AI systems can engage in what she calls “moral self-correction.” In a 2023 paper, she and co-author Deep Ganguli showed that sufficiently advanced models trained with reinforcement learning from human feedback can reduce biased or harmful outputs when prompted with natural-language instructions.

Balancing optimism and risk

Askell operates in an industry under intense scrutiny. AI systems have been linked to misinformation, emotional manipulation and harmful responses in sensitive situations. Public concern about job losses and societal disruption continues to grow.

Yet Askell maintains cautious optimism. She argues that societal “checks and balances” can help keep AI systems under control and believes that how humans treat AI systems will influence what they become.

A philosopher with a conscience

Beyond her technical work, Askell is deeply committed to ethical living. A member of Giving What We Can, she has pledged to donate at least 10% — and potentially more than 50% — of her lifetime income to charity, primarily focused on global poverty.

In 2024, she was named to the TIME100 AI list, recognizing her influence in shaping the future of artificial intelligence.

As AI systems grow more powerful, the question of their character becomes more urgent. At Anthropic, that responsibility rests heavily on Askell — a philosopher tasked not just with building smarter machines, but with shaping their moral compass.