For everyone
Claude
A thoughtful, capable AI assistant. Available on the web, on macOS and Windows, and on phones.
Try Claude →
An AI safety company
Making AI systems
you can rely on.
Anthropic is an AI safety company. We build Claude — a family of large language models — and we do research on how to make them interpretable, steerable, and worth trusting at the scale they will be used.
Read about our approach →Claude 4.7 · Now available
The most capable model we've shipped, with the most careful evaluation work behind it.
Products
All →For developers
Claude API
Build with the same model family that powers Claude. Tool use, prompt caching, batch, and the model context protocol.
Read the docs →
For engineers
Claude Code
A coding agent that lives in the terminal and your IDE. It reads, writes, and ships.
Install →
How we work
Three things that don't change.
01
Safety as science.
We treat the question of how to build safe AI as an empirical one. We hire researchers who would publish their results regardless of where they worked.
02
A small number of bets.
We do a small number of things. One model family, one product, one research agenda. The pace of progress in AI rewards focus.
03
Honest about uncertainty.
Our public communication about model capabilities should be calibrated to what we actually know. Overclaiming, in either direction, is its own failure mode.
Recent research
All research →- May 14, 2026Paper→
On the discoverability of latent capabilities
How small changes to fine-tuning surface abilities that pre-training already had.
- April 02, 2026Paper→
Faithful chains of reasoning
An honest model's chain of thought should reflect what the model actually computed.
- March 18, 2026Policy→
Anthropic's responsible scaling policy, v2.4
Updated commitments on deployment, evaluation, and the conditions under which we slow down.
- February 27, 2026Note→
Specifying what Claude is for
A note on character, tone, and the long work of writing a model's values down.
Our mission, in one sentence
We want the long-term effects of advanced AI to be safe and beneficial. Saying it is the easy part.
Working at Anthropic
We're hiring researchers,
engineers, and a few generalists.
About 612 people across San Francisco, London, Zürich, and remote. We hire slowly. Engineering offers expect a take-home and three technical interviews; research offers expect a published record you'd be proud to defend.
Research ScientistSoftware Engineer, InferencePolicy LeadSafety Engineer
All 87 roles →© 2026 — Anthropic PBC
← all demos