Thomson Reuters on Building Trust in Legal AI | OpenAI

**TL;DR:** Thomson Reuters shares how it uses ChatGPT Enterprise to create a centralized, collaborative legal workspace — boosting drafting efficiency while ensuring AI outputs are verifiable and citable, in response to the legal industry's stringent demands for trust. ## The Daily Grind of Legal Professionals Wait — can we do this differently? It's a thought that recurs constantly as lawyers draft documents. The reality is that lawyers must jump back and forth between different systems while drafting. On top of that, the complexity of drafting global contracts means coordinating with different offices, different colleagues, and even internal and external counsel. All this back-and-forth is pulling lawyers away from the work that truly creates value — the work that demands professional judgment and genuinely benefits clients is being squeezed out by fragmented tools and processes. ## Moving Toward a Centralized Workspace Thomson Reuters' approach is to consolidate collaboration into a single unified environment: ChatGPT Enterprise. The goal is straightforward: communication, drafting, and information retrieval should all happen in one place, rather than scattered across multiple systems. Lawyers no longer need to jump between tools just to find a document, verify a statement, or coordinate with a colleague — they can advance their work within one centralized environment. ## Building on Existing Trust Introducing generative AI doesn't mean rebuilding trust from scratch. Thomson Reuters anchors it in the trust it has already earned through Counsel Legal and HiQu — products that have long established credibility and habitual use among legal professionals. On this foundation, the key value AI brings is this: **judgments become verifiable and citable**. Legal work has no room for answers that merely "sound right" — every conclusion must be traceable, testable, and capable of being put into formal documents with someone's name attached to it. ## A Fiduciary Duty to Legal Professionals When describing this work, Thomson Reuters used a weighty term: **fiduciary duty**. They believe they owe a fiduciary duty to legal professionals, and therefore must deliver information that those professionals can trust and are willing to sign their names to. In other words, AI doesn't output a replacement for advice — it outputs content that a lawyer can personally stand behind and take responsibility for. This sets the bar for the product in terms of accuracy, traceability, and diligence. ## In Summary Thomson Reuters' approach can be distilled into three points: - **Reduce context-switching**: Consolidate collaboration that's currently scattered across multiple systems, offices, and internal/external counsel into a single environment with ChatGPT Enterprise; - **Leverage existing trust**: Build on the industry credibility already established by Counsel Legal and HiQu, rather than asking users to evaluate an entirely new brand; - **Own the results**: Make AI judgments verifiable and citable, meeting the fiduciary standard that lawyers are willing to put their names to. For any AI implementation targeting heavily regulated, high-accountability industries, this philosophy of "credibility over efficiency" is well worth taking to heart. Source: Thomson Reuters on Building Trust in Legal AI | OpenAI (https://www.youtube.com/watch?v=ExVDpMeaWp0)