Thomson Reuters Bets $40M on Proprietary AI as DeepMind Alumni Launch Research Agent Faraday

Legal and scientific sectors see a push toward specialized, in-house AI models over relying on big labs like OpenAI and Anthropic

edit
By LineZotpaper
Published
Updated
Read Time3 min
Sources3 outlets
Thomson Reuters is building its own language model called 'Thomson,' based on Alibaba's Qwen architecture, investing approximately $40 million over two years to own its AI rather than rent from OpenAI or Anthropic. The move comes as Inherent, a British AI lab founded by DeepMind alumni, releases Faraday, an AI agent that claims to outperform rivals at replicating scientific research, signaling a broader trend toward domain-specific AI in enterprise and science.

Thomson Reuters goes its own way

Thomson Reuters is launching 'Thomson,' a proprietary large language model built on Alibaba's Qwen framework, according to a report from The Decoder. The company's CTO Joel Hron emphasized the strategic rationale: 'What matters isn't intelligence itself, but knowing which intelligence you need to own.' The model, costing roughly $40 million over two years, is designed to integrate deeply with Thomson Reuters' own content, such as Westlaw legal databases. Benchmarks show top marks only when the model can tap into that exclusive data, suggesting the company's competitive edge lies more in its curated information than in raw model capability.

Inherent's research teammate

Separately, Inherent, a London-based startup founded by DeepMind alumni, has unveiled 'Faraday,' an AI agent specialized in replicating scientific research. According to TechCrunch, Faraday outperforms models from Anthropic and OpenAI in reproducing experimental results from published papers. Inherent positions Faraday as a 'teammate' for scientists, automating the verification and extension of existing work. The agent uses a combination of retrieval-augmented generation and code execution to simulate experiments.

A shift toward ownership

Both developments reflect a growing skepticism among businesses and research institutions about relying on general-purpose AI from dominant providers. By owning the model and the data pipeline, organizations can tailor capabilities, control costs, and protect proprietary knowledge. Thomson Reuters' approach also hedges against potential price hikes or policy changes from companies like OpenAI and Anthropic. However, the $40 million investment is significant and may not be feasible for smaller firms. Inherent, by contrast, aims to offer Faraday as a service, but its success depends on the accuracy and reproducibility of its outputs—a perennial challenge in AI-driven science.

Critics warn about reproducibility and lock-in

Some researchers caution that AI agents like Faraday may overstate their ability to replicate human-led experiments, as subtle contextual details are often lost. Meanwhile, the move to build proprietary models risks creating new data silos, making cross-institutional collaboration harder. Thomson Reuters' reliance on Alibaba's Qwen also raises questions about geopolitical dependencies, especially as US-China tech tensions persist.

§

Analysis

Why This Matters

  • Organizations are increasingly choosing to build or customize their own AI rather than relying on third-party APIs, redistributing power in the AI ecosystem.
  • Inherent's Faraday could accelerate scientific discovery by automating replication, but its reliability remains unproven at scale.
  • The $40 million investment by Thomson Reuters signals that domain-specific AI can deliver a competitive moat, potentially reshaping how legal and professional services use AI.

Background

Thomson Reuters has long provided legal and business information through Westlaw and other platforms. As generative AI emerged, the company experimented with integrating models from OpenAI and Anthropic but grew concerned about cost, control, and data privacy. The decision to build Thomson on Qwen is a bet on open-weight architectures that allow fine-tuning on proprietary data.

Inherent was founded in 2024 by former DeepMind researchers who specialized in meta-learning and scientific reasoning. Faraday represents their first commercial product, targeting academic and corporate R&D labs. The agent builds on earlier work by DeepMind and others on 'AI scientists' that can read and conduct experiments.

The broader context is a wave of enterprise AI investments. According to industry estimates, the number of proprietary large language models deployed by Fortune 500 companies doubled in 2026. Yet most remain smaller and less capable than frontier models, making domain specialization critical.

Key Perspectives

Thomson Reuters: Believes owning the model is essential for protecting its content moat and delivering reliable AI tools to legal professionals. Inherent: Faraday's creators argue that open replication of research is a major bottleneck and that their agent can significantly speed up validation. Critics/Skeptics: Skeptics question whether Faraday's published benchmarks translate to real-world scientific practice, and note that Thomson Reuters' model only excels when paired with its own proprietary data, limiting generalizability.

What to Watch

  • Adoption of Thomson Reuters' Thomson model by law firms and whether it can lower costs compared to API-based alternatives.
  • Independent replication studies of Faraday's results, ideally by academic labs.
  • Whether OpenAI or Anthropic adjust pricing or terms in response to enterprise flight to proprietary models.
  • Any export control or licensing issues surrounding use of Alibaba's Qwen by a US-headquartered company.

Sources

newspaper

Zotpaper

Articles published under the Zotpaper byline are synthesized from multiple source publications by our AI editor and reviewed by our editorial process. Each story combines reporting from credible outlets to give readers a balanced, comprehensive view.