Thomson Reuters has introduced Thomson, its first proprietary large language model developed in-house, designed to be cost-efficient and fully controlled by the company. Unlike typical frontier AI models that require billions in compute and infrastructure, Thomson was built on a strong open-source foundation with a $40 million investment, leveraging decades of proprietary content and expertise. The model excels in professional tasks, especially legal work, and is already integrated into CoCounsel Legal for structured document review. Thomson Reuters emphasizes AI sovereignty, trust, and accuracy, aiming to set a new standard with Fiduciary-Grade AI™ for professionals requiring high accountability and data privacy. The company plans to expand Thomson's use across its legal and tax services with more sovereign AI options.
