Independent fan archive
Dario Amodei: essays, research and ideas on building safer AI
Dario Amodei is the co-founder and CEO of Anthropic, the company behind Claude. This archive collects his most important essays and papers, explains the ideas behind them in plain language, and links every claim back to its source.
At a glance
Start here
Six ways into the archive, depending on what you want to know.
About Dario Amodei
Background, career path from physics to Baidu, Google Brain, OpenAI and Anthropic.
Read more →
Reading listEssays and papers
A reading list grouped by theme: his own essays, scaling research, safety research and Anthropic work.
Read more →
ExplainerKey ideas explained
Scaling laws, Constitutional AI, Responsible Scaling Policy, interpretability and more, in plain English.
Read more →
ProductThe Claude model family
How Claude went from a 2023 launch to a family of models used for writing, coding and research.
Read more →
ChronologyTimeline
From a 2016 safety paper to today, with dates and sources.
Read more →
AnswersFAQ
Quick answers: who founded Anthropic, what is the RSP, what should I read first?
Read more →
Essays worth reading first
Machines of Loving Grace
His 2024 essay on the upside of powerful AI: biology, neuroscience, development, governance and meaning.
Read more →
Essay guideThe Urgency of Interpretability
Why understanding what happens inside AI models matters, and why the clock is ticking.
Read more →
PaperConcrete Problems in AI Safety
The 2016 paper that gave safety researchers a practical agenda.
Read more →
Deep dives
Focused explainers on the ideas and organisations behind the essays.
Anthropic explained
Founders, structure, strategy, research and products.
Read more →
PolicyResponsible Scaling Policy
AI Safety Levels and how safeguards scale with capability.
Read more →
MethodConstitutional AI
Training a model with written principles and AI feedback.
Read more →
ResearchScaling laws
Why more compute reliably buys better models.
Read more →
Essay guideOn DeepSeek and Export Controls
His 2025 essay on AI chips and geopolitics.
Read more →
ReferenceAI safety glossary
Twenty-one terms in plain English.
Read more →
New here? Follow a guided route on Start here.
Why this archive exists
Few people have been present at as many turning points in modern AI as Dario Amodei: speech recognition at Baidu, early safety research at Google Brain, the scaling era at OpenAI, and now the leadership of a frontier lab that publishes openly about risk. Understanding his work is a useful shortcut to understanding the debates that shape the field.
This site is an admiration project, but it holds itself to editorial standards. Every page separates what Dario has written himself, what Anthropic has published as an organisation, and what is interpretation. Where something is uncertain, it says so.
Looking for the official site? This is an independent fan project. Dario Amodei's own essays are published at darioamodei.com and Anthropic's at anthropic.com.