Evidence item

The paper documents internal frontier-model use and explicitly raises concentration-of-power concerns, supporting the direction of the claim; it does not test whether this trend is the opposite of a Coasean prediction or prove that internal models cause larger firms. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 12450

Source: What Should Frontier AI Developers Disclose About Internal Deployments? · locator {} · excerpt hash sha256:a97ee0499dc3a1e2df31ff7355a802f39d46d372afd8c758c8505ad64c65ef69

Frontier AI developers are increasingly relying on highly capable models for machine learning engineering and AI research and development internal to AI labs. ... Internal deployments also raise concerns about concentration of power within labs or misuse of highly capable internal systems by lab insiders without external oversight.