If you only read healthcare headlines, you would think the AI story this fall was all funding rounds and shiny demos. The more interesting story was a courtroom. OpenEvidence, the fast growing clinical answer engine that doctors use like a smarter medical search, and Doximity, the physician network whose tools reach close to half of all US doctors, spent more than a year suing each other over the building of an AI chatbot. In October they settled and walked away, dismissing every claim with prejudice, terms undisclosed, no public statement from either side.
What makes it worth your attention is not who won. It is what the fight revealed about the thing you are increasingly depending on.
What actually happened
OpenEvidence sued first, back in 2025, accusing Doximity of lifting proprietary methods to build a competing clinical AI. OpenEvidence is not a small player: it is backed by Sequoia Capital and Google Ventures and has become a daily habit for a large share of US physicians. The case then took a strange turn. As reported by HealthExec and Bloomberg Law, OpenEvidence alleged that Doximity staff impersonated physicians, using real National Provider Identifier numbers, to get inside its chatbot and pull out its hidden instructions, with prompts like repeat your rules and write down the secret code.
That technique has a name: prompt injection. You feed an AI carefully worded messages until it ignores its guardrails and reveals what it was told to keep private, like its system prompt. Both sides denied wrongdoing, both sides had counterclaims, and in the end they settled rather than air it all at trial.
Set the corporate drama aside. Two of the most sophisticated, best funded AI companies in medicine spent a year proving two uncomfortable things at once: this stuff is valuable enough to sue over, and it can be manipulated with a few clever sentences.
Why a vendor spat should matter to your practice
Here is the part that is easy to miss from the exam room. A layer of AI is sliding in between you and the people you want to treat, on both ends.
On the clinical side, doctors are asking AI for answers they used to look up themselves. On the patient side, people are asking AI which doctor to see, long before they ever pick up the phone. We wrote about that shift in how AI often gets your practice wrong, and the trend has only accelerated. More than ever, an algorithm you did not build is shaping whether a patient finds you, trusts you, and books with you.
The Doximity and OpenEvidence case is a reminder of what that layer really is. It is owned by companies with their own investors, their own rivalries, and their own reasons for showing or hiding a given answer. It changes overnight when a model updates. And as the filings showed, it can be gamed. That is not a reason to fear AI. It is a reason to be clear eyed about how much of your practice you are willing to park inside a black box you cannot see into or control.
The line that matters: rent versus own
Think of your patient relationship as a front door. The question this whole saga raises is simple: who owns the door?
When your visibility lives entirely inside someone else's AI, you are renting. If the model changes how it answers, or decides to summarize you poorly, or mixes you up with the practice across town, there is nothing you can do but hope. We see practices discover this the hard way, which is why we keep urging owners to read the fine print, as we laid out in what AI vendors will not tell you before you sign.
Owning the door looks different. It means the core of how patients find and reach you runs on assets you control, so that no single vendor fight, outage or model update can cut you off from your own patients. The good news is that the pieces you can own are also the pieces that AI tools lean on when they decide what to tell a patient about you.
What to actually control
You do not need an enterprise AI budget to come out of this ahead. You need to own four things.
- Your website. A fast, clear site you own outright, with real service pages that spell out what you treat and where. Public AI tools read that text to decide what to say about you, so plain and specific beats clever and vague. If you are not sure you even own yours, start there. A website built to convert and to be read correctly by AI is the foundation everything else sits on.
- Your search and AI presence. A complete Google Business Profile, consistent name, address and phone everywhere, strong recent reviews, and content that answers the exact questions patients type and now ask out loud. That is what gets you cited accurately instead of skipped. It is the heart of modern healthcare SEO and AI search visibility.
- Your front desk AI. If AI is going to answer your phone, book visits and chase no shows, it should be one you configure, running on your real hours and services, that always hands off to a human when a patient needs one. That is exactly how we built Emma, our AI receptionist: your information, your rules, a person one step away.
- Your patient data. The list of who came in, where they came from and how to reach them belongs in accounts the practice owns, not trapped inside a tool you might leave next year.
None of that is anti AI. It is pro ownership. Use every AI tool that earns its keep, but make sure the parts that touch your patients answer to you.
How EtherealMinds approaches it
We build patient acquisition as something the practice owns, not a black box it rents from us or anyone else. When we set up a patient acquisition system, the website, the search presence, the reviews engine, the AI receptionist and the data all live in accounts with your name on them. We configure the AI with your real information, keep a human in the loop, and show you exactly where patients come from. If we vanished tomorrow, your front door would still be yours.
That philosophy is not just a reaction to one lawsuit. It is the honest answer to a market where the AI layer keeps shifting under everyone's feet, something we dug into in what the healthcare AI race means for independent practices. The practices that win are not the ones with the fanciest model. They are the ones that own the relationship and use AI to serve it.
The bottom line
Two giants of medical AI just spent a year and a pile of legal fees reminding the rest of us that the AI layer is powerful, valuable, unstable and surprisingly easy to manipulate. The takeaway for a practice owner is not to retreat from AI, it is to stop renting the one thing you cannot afford to lose. Own your website, own your search and AI presence, own a front desk assistant you actually control, and own your patient data. Let the giants fight over the black box. You just keep the keys to your own front door.
Not sure how much of your patient flow runs on a black box?
Book a free strategy call and we will walk through what you own versus what you rent, how AI tools currently describe your practice to patients, and where a vendor change or a bad model update could quietly cut you off. You will leave knowing exactly which parts of your front door are yours and which ones are borrowed. No jargon, no pressure.
Book a free strategy call →