Thomson Reuters Launches Fiduciary-Grade AI Framework for Professionals

Thomson Reuters Corporation has introduced a framework for Fiduciary-Grade AI, establishing a higher benchmark for artificial intelligence designed for professionals subject to regulatory oversight and duties of care. The company argues that general-purpose AI is inadequate for high-stakes professional work because reliance on unverified open-web data creates risks, including fabricated citations.

The Fiduciary-Grade AI framework relies on four core design principles. These include grounding outputs in authoritative and curated content, integrating human expertise during the development process, implementing privacy and security by design, and ensuring that reasoning is transparent and verifiable.

Thomson Reuters is applying these principles to its AI assistant, CoCounsel. This integration is intended to ensure that the tool's outputs can be traced and defended under professional scrutiny, moving beyond the limitations of standard AI models.