Anthropic selects Accenture as first embedded AI safety evaluator

Anthropic has selected Accenture as its first embedded AI safety evaluator, marking a concrete step toward implementing CEO Dario Amodei's proposal to slow down AI development. The companies will each invest at least $1 billion over five years, with Accenture's specialist unit, Faculty, embedding staff inside Anthropic with employee-level access to red-team models. While tech figures like Sam Altman and Elon Musk support the safety framework, others like Nvidia's Jensen Huang have dismissed the need for new regulations.

Anthropic-Accenture embedded evaluator partnership

• ▪Anthropic announced on September 18, 2026, that it has selected Accenture as its first embedded evaluator to independently test the safety of its frontier AI models
• ▪Accenture's specialist AI business, Faculty, will lead Anthropic's partnership with Accenture by embedding employees into Anthropic to red-team models, test safeguards, and conduct alignment assessments
• ▪The embedded evaluators from Accenture's specialist AI business, Faculty, will have employee-level access inside Anthropic to monitor operations, verify safety commitments, and report incidents

Bilateral billion-dollar safety investment

• ▪Anthropic and Accenture have each agreed to invest at least $1 billion over the next five years to build capacity for AI safety evaluation
• ▪Following the announcement of Anthropic's partnership with Accenture on September 18, 2026, Accenture shares rose approximately 7% in extended after-hours trading

Amodei AI development slowdown proposal

• ▪Anthropic CEO Dario Amodei published a proposal on September 12, 2026, calling for a coordinated slowdown in AI research and the integration of third-party evaluators
• ▪The first step of Dario Amodei's proposal to slow down AI development, published in an essay on Saturday, September 19, 2026, grants third-party evaluators employee-level access to AI companies to verify safety practices and report incidents

Industry reactions to safety measures

• ▪Tech companies including Meta and Amazon have declined to fully support Dario Amodei's proposal to slow down AI development, arguing instead that AI companies will regulate themselves
• ▪Nvidia CEO Jensen Huang brushed off concerns regarding Dario Amodei's proposal to slow down AI development, arguing that there is no need for new AI regulation

AI catastrophic risk concerns

• ▪OpenAI announced on September 16, 2026, that it would begin publishing regular reports on unexpected or concerning AI model behavior
• ▪AI researchers have warned about the potential for advanced artificial intelligence to cause catastrophic harm, putting Anthropic and OpenAI under intense scrutiny

Debatable claims

• ▪Commercial consulting firms cannot provide truly independent AI safety oversight
• ▪Tech companies should implement a coordinated slowdown in frontier AI development
• ▪AI companies should not directly fund their own safety evaluators