Thomson Reuters Launches Proprietary Thomson-1 AI Model

Thomson Reuters launched Thomson-1, its first proprietary AI model, to decrease financial dependence on costly third-party services such as Anthropic's Claude. The new model is based on Snowdon, which the company developed by adapting an open-source Qwen model created by the Chinese firm Alibaba.

Chief Technology Officer Joel Hron stated that Thomson-1 will initially focus on document review. The company intends for the model to power an increasing number of capabilities within CoCounsel, its AI legal assistant, over time. To ensure the model was safe and de-biased, the company spent several months collaborating with Imperial College London during development.

The move follows a trend of Western companies using affordable Chinese open-source models to build their own intellectual property and avoid the high costs of renting AI infrastructure. This development occurs as Anthropic has called for restrictions on Chinese AI labs.