What an AI Certificate Certifies, and What It Does Not
![empty formal interior, natural lighting through tall windows, wood paneling, institutional architecture, sense of history and permanence, marble columns, high ceilings, formal furniture, muted palette, A long, polished mahogany table in a cavernous boardroom, its surface scattered with bound reports stamped "AI Governance Framework" and "ISO/IEC 42001", sunlight streaming through tall, arched windows at a low angle, casting deep, sharp shadows across the empty chairs, the air still and heavy with dust motes suspended in the light, the scale of the room dwarfing the documents that bear the weight of certification [fal-ai/z-image/turbo] empty formal interior, natural lighting through tall windows, wood paneling, institutional architecture, sense of history and permanence, marble columns, high ceilings, formal furniture, muted palette, A long, polished mahogany table in a cavernous boardroom, its surface scattered with bound reports stamped "AI Governance Framework" and "ISO/IEC 42001", sunlight streaming through tall, arched windows at a low angle, casting deep, sharp shadows across the empty chairs, the air still and heavy with dust motes suspended in the light, the scale of the room dwarfing the documents that bear the weight of certification [fal-ai/z-image/turbo]](https://cdn.digitalrain.dev/thelongview/viral-images/fec9ea86-95b9-4b3f-bf48-5e8293ef134f_viral_2_square.jpg)
Thomson Reuters has extended its ISO/IEC 42001:2023 certification to more legal and compliance tools, confirming governance processes are in place—not that outputs are more accurate. The certificate tracks how AI is managed, not whether it gets it right.
Thomson Reuters has widened the scope of its ISO/IEC 42001:2023 certification across a larger set of legal, risk, fraud, tax, audit and accounting products, and the change is best read as an extension of a governance process rather than a new capability. The standard is an international specification for AI management systems, so what it certifies is the way a company develops, reviews, deploys and monitors AI, checked by external auditors, not the answers its tools give. Certification runs on a three-year cycle: an initial audit, annual surveillance audits in years one and two, and recertification in year three. Scope is set product by product. The company says Westlaw became the first certified product in March 2025, after a ten-month process of internal and external audits covering risk management, data governance and Responsible AI practice, and that CoCounsel Legal followed shortly afterwards. The current scope now names Practical Law, CLEAR Investigate, CoCounsel Tax, Audit and Accounting, and Westlaw Advantage. The controls the company describes are procedural: oversight scaled to risk, contractual terms with third parties, staff education on approved tools, additional reviews for higher-risk uses, and feedback from customers, all aligned with the NIST AI Risk Management Framework. Public detail on audit findings, the number of use cases reviewed or measured error rates is not yet available, so the certificate is evidence that governance exists and is audited, not evidence that outputs are more reliable. The company's Fiduciary-Grade label describes an intention that AI should support professionals who remain accountable for their decisions, and it says that almost right is not good enough. The certification does not test that threshold, so the claim that it proves trustworthiness is unsupported by the evidence offered. For a firm weighing these products, the operative question is whether the controls reduce error in a specific workflow, and the certificate applies only to the products named in its scope.
† Analysis: the writer's conclusion from the reported facts, not a fact stated in the source.
- The article infers that the certification should be viewed as an extension of a governance process rather than a new capability, and that it is evidence of audited governance rather than a guarantee of output reliability. These are logical conclusions drawn from the source material's explanation that the certification covers the AI management system and governance framework rather than the specific accuracy of individual tool outputs.
Published November 13, 2026