Anthropic

Anthropic is an AI research and safety company founded in January 2021 by siblings Dario and Daniela Amodei, both former OpenAI executives. It operates as a Public Benefit Corporation and develops the Claude family of large language models, with Claude 3.5 Sonnet released in June 2024 achieving a 92.0% score on the GPQA graduate-level reasoning benchmark and 90.2% on HumanEval coding, per the company’s technical report. The firm introduced its Constitutional AI training methodology in the December 2022 paper 'Constitutional AI: Harmlessness from AI Feedback,' which uses a fixed set of principles to steer model behavior. By early 2024, it had secured a $2 billion investment from Google and a $4 billion commitment from Amazon, with an additional $4 billion option. Anthropic published a formal Responsible Scaling Policy in September 2023, defining risk thresholds and safety protocols. The company matters because it directly challenges OpenAI and Google at the frontier of AI capability while binding its corporate structure and research agenda to long-term safety commitments, actively shaping both market competition and emerging AI regulation.

Two-hop subgraph: this entity, every entity it directly relates to, and every entity those neighbors relate to. Drag a node, scroll to zoom, click to inspect — or click any neighbor and re-center the atlas there.