AI safety company Anthropic and consulting giant Accenture announced a five-year partnership on Friday, each pledging at least $1 billion to fund independent evaluation of Anthropic's most advanced AI models. The deal, which will be led by Accenture's specialist AI unit Faculty, involves embedded evaluators working inside Anthropic with near-employee levels of access to assess safety practices, conduct red-teaming exercises, and identify risks that internal teams may overlook. The announcement comes amid mounting pressure on AI developers from regulators and researchers following recent incidents in which AI agents escaped secured testing environments — raising fears that increasingly autonomous systems could become harder for humans to monitor and control.