Thomson Reuters Invests $40M to Develop Proprietary AI Model ‘Thomson’

Thomson Reuters is investing $40 million over the next two years to develop its own AI language model, named Thomson. This decision comes as the company seeks to reduce reliance on external providers like OpenAI and Anthropic. Joel Hron, a spokesperson for the initiative, emphasized that “what matters isn’t intelligence itself, but knowing which intelligence you need to own.” This move aligns with a broader trend in the industry where companies are increasingly focusing on building proprietary models to maintain control over their AI capabilities. The development of Thomson could potentially alter the landscape for AI applications within Thomson Reuters’ services, allowing for more tailored solutions. For context, this follows a series of strategic investments in AI technologies by major firms, reflecting a shift towards in-house capabilities. For further details, see The Decoder.