OpenEvidence, a medical AI platform used daily by more than 40 per cent of physicians across 10,000 hospitals in the United States, has withdrawn its clinical decision support application from the UK and European markets. The company cited the regulatory landscape as the primary reason for the exit, specifically referencing the uncertainty generated by the EU Artificial Intelligence Act and its implications for tools that influence clinical decision-making.
The withdrawal carries financial context that makes it notable. OpenEvidence raised $250 million in January at a valuation of $12 billion, making it one of the most highly capitalised AI health companies in the world. Its platform supported an estimated 18 million clinical consultations in a single month in the United States, a scale of deployment that places it among the most widely used medical AI tools currently in operation. The decision to exit two major markets is not, therefore, a retreat by a company struggling for commercial traction. It reflects a calculated assessment that the legal risks of operating under the current European framework outweigh the near-term commercial opportunity.
The EU Artificial Intelligence Act classifies tools that support clinical diagnoses as high-risk AI systems. That classification triggers requirements for transparency in how the system reaches its outputs, clinical validation evidence, and clarity on liability where the tool contributes to a clinical decision that results in patient harm. Those requirements are not inherently unreasonable as a matter of patient safety policy, but the detailed compliance obligations and the unresolved questions around liability allocation have created a degree of legal uncertainty that companies developing in this category are finding difficult to price and plan around.
The UK, while no longer subject to EU legislation directly, has been monitoring the AI Act closely and has not yet established an equivalent statutory framework of its own. The resulting position is one in which UK regulators have issued guidance and principles without binding legislation, leaving companies uncertain about the legal basis on which high-risk medical AI can be deployed, validated, and held accountable. For a company with the legal and financial resources of OpenEvidence, that ambiguity is apparently sufficient to make market entry unattractive relative to the comparatively settled US environment.
The practical consequences for UK clinicians are direct. Doctors who would have had access to a platform handling millions of evidence-based clinical queries monthly will instead rely on the tools currently available within the NHS digital ecosystem, which do not replicate that functionality at equivalent scale or sophistication. Health systems that had incorporated OpenEvidence into planned digital transformation programmes will need to revise those plans, either identifying alternative platforms or deferring the relevant workstreams.
The withdrawal also carries implications for the government's stated ambitions in health technology. Ministers have repeatedly described AI adoption in the NHS as a priority, and the NHS has published frameworks for accelerating the integration of clinical AI tools. The exit of a platform of this scale from the UK market, for reasons directly related to regulatory uncertainty, sits in tension with those ambitions and raises questions about whether the current regulatory approach is aligned with the pace of deployment the government says it wants to achieve.
The counter-argument to that concern is the one that motivated the EU AI Act in the first place. Clinical decision support tools operating at scale in high-stakes medical environments carry genuine risks if they are poorly validated, opaque in their reasoning, or deployed in contexts for which they were not designed. Regulatory caution in this domain is not without justification, and the consequences of moving too quickly are not trivial. The question is whether the current framework has found the right balance between those competing concerns, or whether it has set compliance requirements at a level that excludes well-resourced and clinically validated platforms alongside genuinely problematic ones.
OpenEvidence's withdrawal does not resolve that question, but it provides a concrete data point. A company with substantial capital, clinical validation evidence from its US deployment, and a demonstrated user base has concluded that the European regulatory environment is currently too uncertain to enter. Whether other companies in the same category are reaching similar conclusions, and whether that pattern is sufficient to shift the policy debate in Westminster or Brussels, will become clearer as further deployment decisions are made in the months ahead.
