Anthropic's first outside safety checker is a consulting firm.
Anthropic said on Friday that staff from Accenture will work inside the company as embedded evaluators. They get employee-level access to training runs, deployment decisions and the people building the models. Each side plans to spend at least $1 billion over five years, and more evaluators are coming.
Why it mattersLabs have graded their own safety homework for years. This is the first deal that lets outsiders watch a model before it ships.
▶ Related video: Anthropic CEO calls for competitors to agree to "third-party evaluators" in push for AI safety
Read in the app — free, in 9 languages
Related stories
Canada and Germany put C$300m behind an AI built to have no goals.2026-09-19Anthropic leased part of a A$32bn data centre in outback Queensland.
2026-09-19A startup that insures AI agents has raised $55 million.
2026-09-15Nvidia and Palantir cut back on Claude over a 30-day log rule.
2026-09-15Superhuman bought the AI notetaker that sits in your meetings.
2026-09-15