Public benefit corporation building Claude with a stated focus on AI safety and a Long-Term Benefit Trust.
Anthropic builds Claude and markets itself as the safety-first lab. Its record is mixed: it held a costly red line against the Pentagon but also softened its own scaling policy and paid a record copyright settlement.
Focus
- Safety evaluations
- Alignment research
- Transparency
- Human oversight
Funding, as far as we know
Corporate: Amazon (~$8B), Google (~$3B), plus venture rounds - https://www.anthropic.com/news
What they've actually done
Activated ASL-3 safeguards (CBRN classifiers, weight-security controls) for Claude Opus 4
22 May 2025
Agreed to pay $1.5B to authors for pirated training books; final approval July 2026
2025-09 / 2026-07-22
Released RSP v3.0, removing the explicit pause commitment in favour of Risk Reports and a roadmap
24 February 2026
Refused the Pentagon's demand to lift weapons/surveillance restrictions, was blacklisted, and won a ruling that the blacklist was unlawful retaliation
2026-02-26 / 2026-08-28