Thomson Reuters has introduced an AI platform for tax, audit and accounting work that makes data accuracy and transparency its key selling point. Launched on 22 October, the new "CoCounsel" system is designed to deliver answers that are backed up by source documents and customers’ own data ([1]). The company says this next-generation tool can provide “finished, defensible” results that professionals can review and trust ([2]), addressing the well-known problem of AI models generating plausible-sounding but incorrect information. The launch comes as research shows 43% of data and analytics leaders see lack of data readiness as their biggest barrier to aligning AI with business strategy ([3]). Thomson Reuters’ move suggests that for high-stakes enterprise applications, success with AI depends on getting the underlying data right and ensuring every output can be traced back to a reliable source.
In the healthcare sector, Germany’s Boehringer Ingelheim and Microsoft announced a strategic collaboration on 22 October to harness real-world clinical data for AI-driven drug development ([1]). The partnership will focus on creating "Digital Patient Twins", virtual models of patients that use existing clinical and trial data to predict disease progression and treatment outcomes ([2]). By leveraging historical patient data and AI, the companies aim to reduce the number of patients needed in control groups and accelerate the generation of clinical evidence in trials ([3]). For data and technology leaders, this highlights how combining domain-specific data with cloud and AI expertise can boost innovation. It also reflects a growing trend of organisations pooling their data and capabilities to overcome challenges in training advanced AI systems.
Meanwhile in Washington, US lawmakers are responding to concerns about the security of AI training data. On 22 October, a bipartisan group in Congress introduced the "AI Cyber Defence Act" ([1]). The bill would direct the National Institute of Standards and Technology (NIST) to develop voluntary standards to thwart "distillation attacks", techniques by which foreign adversaries from countries like China and Russia might steal proprietary AI model data by querying the models ([2]). One sponsor of the bill warned that it is “incredibly expensive” for US companies to stay even 6, 12 months ahead in AI, and that allowing rivals to copy models undermines innovation ([3]). The legislative push underscores how critical data and model protection has become in the AI race. For leaders, it is a reminder that data readiness now includes safeguarding the data and intellectual property that give your AI systems a competitive edge.