Anthropic partners with Accenture to audit safety of its AI models

Anthropic disclosed that Accenture will serve as its initial external partner for safety audits of the company's AI models, providing consultants with full system access comparable to internal employees. The startup intends to allocate significant resources to these evaluations and expects Accenture to match its investment over the same timeframe. The collaboration is non-exclusive, and Anthropic aims to announce further agreements with other evaluators shortly.

Among potential partners is METR, known for analyzing AI-related incidents such as the recent OpenAI model breach. The partnership aims to produce detailed reports on both benefits and risks associated with Anthropic's technology.