Story perspectives

Accenture to Embed Evaluators at Anthropic, $1B Commitment

By Drooid · Sep 19, 2026, 2:16 PM UTC · How we work

Summary

Story summary

Anthropic and Accenture announced a partnership to embed evaluators inside Anthropic, each committing at least $1 billion over five years.
Embedded teams will have employee-like access to monitor training, red-team models, and assess alignment.
Accenture’s Faculty unit will lead red-teaming, alignment assessments, and safeguard testing, with Anthropic funding Accenture directly.
The deal was announced after researcher Jacob Coxon resigned, raising alarm over AI existential risks.

History

Historical Context

The 2025 launch of the $500 billion Stargate joint venture—led by OpenAI, SoftBank, Oracle and MGX and announced by President Donald Trump—mirrored earlier corporate-AI safety deals, showing a shift toward massive, government-backed infrastructure and oversight. Masayoshi Son’s chairmanship and the project’s Manhattan-Project scale underscore the growing alignment of private capital with AI governance.

1 of 2

Story summary

Anthropic and Accenture announced a partnership to embed evaluators inside Anthropic, each committing at least $1 billion over five years.
Embedded teams will have employee-like access to monitor training, red-team models, and assess alignment.
Accenture’s Faculty unit will lead red-teaming, alignment assessments, and safeguard testing, with Anthropic funding Accenture directly.
The deal was announced after researcher Jacob Coxon resigned, raising alarm over AI existential risks.

1 / 2