In June 2025, OpenEvidence filed a lawsuit in the US District Court for the Northern District of California against Veracity Health Inc., the corporate entity behind Vera Health, and the company's co-founder Taieb Bennani. The case was settled several months later, on terms that were not made public. This article sets out what is actually known about the case, is careful throughout to distinguish allegation from established fact, and avoids characterising either company as having behaved dishonestly.
What OpenEvidence alleged
According to the publicly filed complaint, OpenEvidence alleged trade secret misappropriation and tortious interference, among other claims. The complaint reportedly described more than 600 instances of what it characterised as sophisticated attempts to access OpenEvidence's platform by impersonating healthcare professionals, and alleged that credentials obtained this way were used to extract training-related material across multiple medical specialties, subsequently used, OpenEvidence claimed, in the development of a competing product. These are allegations made by one party in a legal filing. They were not tested at trial, and no court made a finding on whether they were true.
What prompt injection means in this context
Prompt injection refers to attempts to manipulate an AI system, through carefully constructed inputs, into behaving in ways its developers did not intend, including revealing internal instructions or configuration that the system's operator considers proprietary. It is a real and actively discussed category of technique in AI security more broadly, and OpenEvidence has separately pursued similar claims against other companies, making this litigation part of a wider pattern of disputes over how competitors are permitted to interact with each other's AI systems during this period, rather than an isolated incident specific to Vera Health alone.
How the case concluded
Court filings indicate that OpenEvidence and Veracity Health reached a settlement, reported to the court on 31 October 2025, with the parties indicating they expected to finalise terms within roughly 30 days. The specific terms of that settlement were not made public. A settlement of this kind is not a legal finding that the underlying allegations were true, and it is equally not a finding that they were false; parties settle litigation for many reasons, including the cost and disruption of continued proceedings, entirely independent of the ultimate merits of the claims involved.
Why this dispute matters beyond the two companies directly involved
Regardless of its specific outcome, the case touched on several genuinely unresolved legal questions relevant to the wider AI industry. Whether an AI system's internal instructions, its system prompt, can be protected as a trade secret in the same way traditional software code has long been treated. Where legitimate competitive product testing ends and unlawful extraction of proprietary material begins, a genuinely difficult line to draw with any AI system that can be queried freely by anyone who can obtain access. How competitors in this fast-moving sector may permissibly benchmark and evaluate each other's products. And what obligations follow from using verified professional credentials, of the kind healthcare platforms rely on for access control, to gain entry to a competitor's restricted system.
Maintaining strict neutrality throughout
It is worth restating plainly: the allegations described above were made by OpenEvidence in a legal filing. Vera Health and its co-founder were named as defendants, not found liable, and the case's resolution through settlement, rather than a trial verdict, means no court determination exists either confirming or rejecting the specific factual claims at issue. Neither company should be characterised, on the basis of this litigation alone, as having acted dishonestly, and this article deliberately avoids doing so in either direction.
Connecting this to wider clinical AI competition
This dispute sits within a broader pattern across the clinical AI sector: a relatively small number of well-funded, structurally similar products competing for the same clinician attention, with visible differentiation between them sometimes limited, and proprietary retrieval, ranking and grading systems increasingly treated by their developers as core, defensible competitive assets rather than incidental technical detail. As this category continues to mature and attract further investment, similar disputes over what counts as legitimate competitive scrutiny versus unlawful extraction seem a plausible continuing feature of the space, not a one-off event specific to these two companies.
Where iatroX sits in relation to this
The most defensible competitive position in a sector where disputes of this kind are becoming more common is transparent methodology, genuinely original product development, and openness to independent clinical evaluation of the kind this content series has repeatedly called for across every platform it has examined, iatroX included. This litigation is not, and should not be treated as, a basis for a direct sales argument in either direction; it is a reminder that the underlying legal and competitive terrain around proprietary AI systems in healthcare remains genuinely unsettled, and worth watching with the same neutral scrutiny this article has aimed to apply throughout.
A note on keeping this article current
Given that settlement terms were not disclosed and further developments in AI-sector litigation of this kind are likely, this article reflects what was publicly known at the time of writing and should be revisited if further substantive information becomes available.
Explore iatroX's approach to transparent, UK-grounded evidence →