Thomson Reuters has launched a proprietary large language model that the company developed in-house and trained on its decades of content from Westlaw, Practical Law, Checkpoint and Reuters, the company said in a Monday (Aug. 24) press release.

Dubbed “Thomson,” the LLM began with an open-source foundation and was trained on the company’s proprietary content and domain expertise, according to the release.

As a result, Thomson is fully controlled by Thomson Reuters and does not have the inference costs of typical frontier models, the release said.

In addition, Thomson Reuters designed the LLM for professionals with duties of care and accountability, and the company does not use customer data to train the model without explicit consent, per the release.

Thomson Reuters CEO Steve Hasker said in the release that the company is initially putting Thomson to work in CoCounsel Legal and plans to keep adding capabilities and sovereign AI options. The company said in the release that its plans include capabilities across the legal and tax portfolio.

“Thomson proves what’s possible when you build AI on decades of proprietary content and editorial expertise,” Hasker said. “That’s an advantage only Thomson Reuters has, and it shows in the results: our early evaluations put Thomson on par with the latest frontier models across a range of tasks.”

Thomson Reuters Chief Technology Officer Joel Hron said in the release that Thomson shows that there is an alternative to “bigger models, more compute, more money.”

“Start with a strong foundation, specialize it deeply for the work that matters, and you can build intelligence that is highly capable, far more efficient and entirely under your control,” Hron said.

Source link