Episode 372: Who Controls Your Health Data? Palantir, the NHS and the Risks of Digital Dependency

In this episode of The International Risk Podcast, Dominic Bowen speaks with Phil Booth, coordinator of medConfidential and a long-standing campaigner on medical confidentiality, patient consent and data governance, about what Palantir’s growing role in the NHS reveals about public trust, private technology companies and the data infrastructure increasingly underpinning the modern state.

The conversation examines the NHS Federated Data Platform, the use of Palantir Foundry and the wider risks that arise when critical public infrastructure becomes dependent on private technology companies. Phil argues that the central issue is not only whether the software works, but who controls it, how easily it can be scrutinised or replaced, and whether patients have any meaningful choice over how their health data is used.

Dominic and Phil discuss the limits of pseudonymisation, weaknesses in current opt-out arrangements, the commercial value created around NHS workflows and data systems, and the danger of long-term vendor lock-in. Phil reflects on earlier disputes surrounding care.data and the extraction of GP records, arguing that successive governments have repeatedly failed to treat public consent as a necessary condition of legitimate health-data use. They also explore how Palantir’s work with military, intelligence and policing organisations can create ethical and strategic tensions when the same company becomes deeply embedded in healthcare systems.

The International Risk Podcast brings you conversations with global experts, frontline practitioners, and senior decision-makers who are shaping how we understand and respond to international risk. From geopolitical instability and organised crime to cybersecurity threats and hybrid warfare, each episode explores the forces transforming our world and what smart leaders must do to navigate them. Whether you’re a board member, policymaker, or risk professional, The International Risk Podcast delivers actionable insights, sharp analysis, and real-world stories that matter.

Dominic Bowen is the host of The International Risk Podcast and Europe’s leading expert on international risk and crisis management. As Head of Strategic Advisory and Partner at one of Europe’s leading risk management consulting firms, Dominic advises CEOs, boards, and senior executives across the continent on how to prepare for uncertainty and act with intent. He has spent decades working in war zones, advising multinational companies, and supporting Europe’s business leaders. Dominic is the go-to business advisor for leaders navigating risk, crisis, and strategy; trusted for his clarity, calmness under pressure, and ability to turn volatility into competitive advantage. Dominic equips today’s business leaders with the insight and confidence to lead through disruption and deliver sustained strategic advantage.

Subscribe for all our updates! 

Tell us what you liked!

Transcript

[00:00:01] Phil Booth: So, unfortunately, while, as you say, all the good things are possible, the way in which successive governments and NHS management have acted over the last decade or so means that we are still in a position where there is no clear choice for someone who says: look, I simply do not want my data flowing through Palantir or the Federated Data Platform, if Palantir weren’t supplying that.

[00:00:26] Intro: Welcome back to The International Risk Podcast, where we discuss the latest world news and significant events that impact businesses and organisations worldwide.

[00:00:34] Dominic Bowen: And I have a quick favour to ask before we start today. If you are a regular listener, please subscribe to and follow The International Risk Podcast. It is the simplest way to support the show, and it helps us reach more listeners who need this content. And my commitment to you is that we will keep improving every part of the experience: the guests, the quality of the research and the practical insights we provide. And if there is a guest you think we should bring on the podcast, or a risk that you want unpacked, send it through to us. I promise we read all of your comments. Please hit the subscribe or follow button now, and let’s jump in to today’s episode.

[00:01:08] Dominic Bowen: Have you ever consented to giving your private health records to Palantir? What about to the founder, Peter Thiel? No. Well, then you should listen to this conversation today. Today on the podcast, we are looking at what happens when data infrastructure is built and maintained by private technology companies. Now, the immediate story is Palantir’s role in the UK’s National Health Service Federated Data Platform. NHS England argues that the platform can help hospitals use their own data more efficiently. It will reduce delays, manage waiting lists, coordinate discharge and improve patient care. These are all good things. It also says that Palantir cannot own, commercialise, or train AI models on NHS data. But for many people, including doctors and patients, that does not end their concerns.

[00:01:56] Dominic Bowen: Health data is among some of the most sensitive and personal information any institution can hold. It reveals our illnesses, the medications we are taking, our mental health status, our reproductive history, our disabilities, our family circumstances and, ultimately, our vulnerabilities. So the question is not just how Palantir’s technology works, but whether the public thoroughly understands it, whether they trust it, whether they consent to it and, ultimately, whether they can opt out if they are not comfortable with it.

[00:02:23] Dominic Bowen: To help us unpack this, I am joined today by Phil Booth. He is the coordinator of medConfidential, and he is one of the UK’s leading civil society voices on medical confidentiality, NHS data governance, patient consent and opt-outs. He is also the national coordinator for NO2ID. Phil has been deeply involved in debates around data sharing, data programmes and the governance of large-scale data systems. I think today’s conversation will be really interesting. Phil, welcome to The International Risk Podcast.

[00:02:54] Phil Booth: Thanks for having me, Dominic.

[00:02:56] Dominic Bowen: Phil, whereabouts in the world do we find you today?

[00:02:58] Phil Booth: I am down in East Sussex, in the lovely sunshine. Or actually hiding from it a bit.

[00:03:03] Dominic Bowen: Hiding from it. I understand the heatwave in the UK is quite severe right now.

[00:03:07] Phil Booth: Yes, a little warm.

[00:03:08] Dominic Bowen: It is a little warm. Well, I am glad to hear you are in the shade right now, but I think we should jump right into the conversation. Phil, can you tell us, from your perspective, what the real problem is here? Is it Palantir specifically? Is it the way the NHS is communicating how it is doing the data sharing? Or is there potentially a bigger failure in how public institutions are working and managing and protecting our sensitive data?

[00:03:31] Phil Booth: Well, I think it is aspects of all three, to be honest. If I approach the problem: Palantir is a problem because of what it does for its customers. We always said, okay, there is going to be software that serves all sorts of customers. But if you are serving, on the one hand, a bunch of customers that are trying to kill people, and on the other hand, customers that are about preserving life and sustaining life, you are going to find some serious conflicts. In the NHS, for example, you might find that people are coming to the NHS for healthcare provided at least partly by the software of a company whose software has been used to kill their relatives.

[00:04:10] Phil Booth: This is obviously a personal, moral and ethical problem, and it is completely detached from whether or not the software itself is competent or competently managed. Palantir itself is not magic. It has certainly got a huge marketing and lobbying budget, and it has managed to convince many governments that it is frankly the only way to do things. But at its heart, certainly Palantir Foundry, the system the NHS is using and which is increasingly part of the Palantir ecosystem, if you like, is essentially Apache Spark open-source software, and could be run by any number of companies in any number of ways.

[00:04:47] Phil Booth: What Palantir does is bring in its forward-deployed engineers, essentially consultants who take on roles and develop systems that do what their customers want. NHS England, which is the customer in this case, has got them to do various things, and that is termed the Federated Data Platform. That is not to say we did not already have data platforms in the NHS. We have, and have had for decades. But the things that Palantir can do in terms of creating pipelines, workflows and what have you are being progressively engineered deeper and deeper into the system.

[00:05:25] Phil Booth: One problem is that this is not just a simple company working for a customer. Palantir has a habit, an approach indeed, of creating, if you like, monopolistic functions built on top of this generic platform. In the NHS, that is a thing called Cancer 360 for cancer treatment and a thing called OPTICA about patient discharge. These things only work on Palantir Foundry. There have been cases, I think, in the United States where a police force found a replacement for Palantir, and Palantir resisted handing over its intellectual property. It may not own the data, but it builds things based on those flows of data and how they are processed, which it then jealously guards and which then make replacing Palantir very much harder.

[00:06:13] Dominic Bowen: You gave an example at the start about some of the risks. We talk about, for example, if Palantir’s software is used by the Israeli military to target someone on its most-wanted list, and then someone who is seeking shelter and refuge legally in the UK goes to hospital, that person could be the brother or relative of someone who was killed by Israel the day before using Palantir technology. That example sounds absolutely shocking, but it makes sense and we can understand how that occurs. But for the average person, so the average UK citizen, or even someone in Sweden or Norway, whose governments might also be looking at this technology, who says: all my family is in Sweden, or all my family is in the UK, we pay our taxes, we are really boring. The standard line is: we have nothing to hide anyway. Why would they be worried? Why should they be concerned or considering the risks of this sort of system?

[00:06:56] Phil Booth: With regard to any system of this type, lots of people may not care about what is in their medical history right now, but they cannot say that they would not find something sensitive at some point in the future. Health has that tendency to throw up unexpected things, and systems of this type allow access to all sorts of people that you are not directly interacting with and that you are not aware of. It is these sorts of things that explain why we have lobbied hard for, won and got various opt-outs for uses beyond people’s actual direct care. The big opt-outs that we have at the moment are to protect one’s GP record. The family doctor in the UK has the most complete medical history — it is not totally complete, but the most complete medical history.

[00:07:43] Phil Booth: That is what everyone is seeking to extract. But that information is obviously vital to the delivery of care to you. That is direct care. But it could also be put to various other, what are called secondary, uses. These are lumped together as things like planning or research, but there is a wide range of uses which, frankly, for decades, the British public, when asked, have said they are uncomfortable about: some uses altogether, like it being used by insurers or being used for commercial gain. So there is clearly a case for large numbers of people to be given an option about what is done beyond their direct care.

[00:08:21] Phil Booth: Indeed, there are even options about how your data is shared for your direct care. That is why we need to have these systems consensual, safe and transparent. You must have the choices. The data must obviously be handled safely. That is not just technically; that is all the rules, the governance and the people who access it. And transparent: people need to be able to see what is being done. Unfortunately, although Palantir Foundry certainly has the capability to deliver transparency about what is being done with the data, it is not doing so. Palantir says that it will do what the law of the particular jurisdiction it is working in requires, and will manifest that in its software. People have the right to make a subject access request about their personal data and what is flowing through Palantir is pseudonymised, but it is personal data. So you need to build a button so that someone can find out who has accessed their data. Palantir is not magic, but it is competent software, and it needs to deliver the things that are required for trust.

[00:09:21] Dominic Bowen: You talk about consent, and you talked about opt-outs. I think most patients understand and accept that their information will be used for direct care, and that is generally seen as a good thing. They expect doctors, nurses and the hospital team to access their data in order to assess how best to treat them. But the NHS’s Federated Data Platform, and health-data platforms used by most countries, are complicated spaces. They can fit into operational planning, waiting lists, discharge coordination and, even more broadly, population health and national analytics. Of course, potentially one day, and that may not be far away, they can also involve AI-enabled decision-making within hospitals and healthcare settings. So do the current consent and opt-out arrangements actually give patients meaningful control over their personal health data? And how do patients actually know how they opt in and opt out?

[00:10:12] Phil Booth: Unfortunately, they do not. Where we are now is a consequence of more than a decade of actions where government has tried to grab the GP data of every patient in England. The first one, notoriously, was called care.data back in 2014. Then there was another attempt in late 2021, certainly in the late pandemic period, called GPDPR. Essentially, the current one is something called the single patient record. The recently resigned Health Secretary, Wes Streeting, in his 10-year plan for the NHS, said: we are going to have a single patient record, and that will involve sucking up all the GP data and making the Secretary of State data controller, rather than your GP, a registered medical professional with duties of care and confidence towards you.

[00:11:01] Phil Booth: So you can see this is a sort of large-scale strategic battle, and during the course of that we have managed to preserve an opt-out at the GP practice level. Where your data is currently held, you can deploy what is called a Type 1 opt-out, which means that the GP practice will not release your data for any purpose other than your direct care. That is, if you like, the backstop. There is something called the national data opt-out, which NHS England offers. You can opt out of planning and research uses of the other data that we have from hospitals, clinics and other parts of the NHS. But unfortunately, in practice, the evidence shows that this is not respected.

[00:11:41] Phil Booth: In all too many cases, they advance the argument: well, we have de-identified it or pseudonymised it, so we will treat it as if it is not personal data. That just is not true. Pseudonymised data that is this rich in detail and that is longitudinal, which is linked over time and across contexts, is inherently identifiable, and therefore personal data. The very fact that they link stuff together between contexts demonstrates that it is identifiable data. We have got a problem in that the government keeps passing laws to try to minimise people’s legal protections. We have strategies and policies of successive governments, and those who are lobbying them in the research sector, for example, who just want to get this fantastically valuable data.

[00:12:30] Phil Booth: One of the largest research studies, called UK Biobank, is not just people’s health records. People gave tissue and stuff. So this is genomic data as well. This stuff is showing up for sale on Alibaba because they have not looked after the dataset in the way that they should. So, unfortunately, while, as you say, all the good things are possible, the way in which successive governments and NHS management have acted over the last decade or so means that we are still in a position where there is no clear choice for someone who says: look, I simply do not want my data flowing through Palantir or the Federated Data Platform, if Palantir were not supplying that.

[00:13:10] Dominic Bowen: I think you are right. Well before large language models and artificial intelligence became all the rage over the last three to four years, there had been countless studies where what seems like perfectly anonymised data is very quickly attributable to individuals. Data scientists are very good at taking anonymised data and using it to identify specific people. I think now, with the increasing computing and intelligence power that we have, that is even easier. Your point is very valid there.

[00:13:35] Dominic Bowen: To circle back to your point about whether it is the police, the military or hospital systems: with the military and the police, I can understand that their missions and objectives are much less concerned with privacy. But when we look at the Ministry of Health, the Department of Health and the National Health Service in the UK, I wonder if you can help us understand this. In the UK, there is a healthy level of debate with the government and within the government. In many other countries, that is not the case. People just accept what the government says. I am currently based in Sweden, and certainly in Sweden, and in many other northern European countries, there is an acceptance that our government is good. Our government will always make the best decisions in our interest. Sadly, we know that is not always the case.

[00:14:16] Dominic Bowen: But in the UK, where there is robust, healthy and generally respectful debate between parties, why would the NHS buy software and build such a dependency on a private platform like Palantir that is so difficult to scrutinise, difficult to replace and difficult for democratic institutions to control? This does not seem to make sense as to why they would make that decision.

[00:14:37] Phil Booth: To look at the decision, you have to maybe look at the context. We have robust debate because successive governments have repeatedly screwed up, and not just in health. On ID, for example, when I was doing NO2ID before, they lost some discs of child benefit data, and that changed the public debate. Everyone suddenly went: hang on, you lost my bank details. Why can I trust you with anything else? So government failure and incompetence has conditioned the debate in the UK for several decades now. Billions, tens of billions, have been spent on all sorts of large-scale systems in the NHS. Something called the National Programme for IT: £12 billion, basically no meaningful result.

[00:15:19] Phil Booth: The reason why, I guess, was that they wanted to buy something that was actually going to work and that was competent. But the reason they made that choice is simply that they had not invested in what Professor Ben Goldacre said in his 2022 review of health data: they had not invested in people and teams and tools and code, so that we could do this ourselves. The Civil Service is one thing, and the NHS is another. There are, if you like, parallel bureaucracies. But the NHS is one of the largest organisations in the world in terms of number of employees, and it should have within it the capability of developing, supporting, creating, modifying and running the software that it needs to.

[00:16:09] Phil Booth: We used to have a separate independent body. It used to be called, I think, the Health and Social Care Information Centre. Then, after care.data, it was renamed NHS Digital. This was where a lot of the central systems — the spine systems of the NHS, if you like — were managed. Well, these are being shifted over to this private provider.

[00:16:28] Phil Booth: This does seem to be a failure of a desire to invest properly in your own national capabilities. Frankly, these things never actually work out in the end being any cheaper. You find all sorts of problems with being locked into the provider’s system, because any provider that does get a foothold in a large-scale system like this is obviously going to try to keep its foothold there. Sorry, that was a bit of a roam around, but you can see there are probably multiple factors that influence the decision. Once that decision has been made, it then influences further decisions.

[00:17:06] Phil Booth: You mentioned AI, for example. NHS England has just announced the NHS AI platform, which, surprise, surprise, is actually just a rebranding of Palantir’s AI platform. That interacts very interestingly with international affairs, because, as listeners may know, Anthropic got into a disagreement with the Secretary of State for Defence, Pete Hegseth, over in the United States. Pete Hegseth said that Palantir, or all government suppliers, had to make sure that Anthropic was not in their supply chains. Well, Anthropic is clearly in Palantir’s supply chain, because it was used for, I think, some of the action in Venezuela, but it is also part of the NHS testbed. So one customer making one decision based on one set of circumstances can potentially affect the choices that Palantir can offer to its other customers.

[00:18:01] Phil Booth: Right now, as complex information processing, large language models and all those sorts of things develop very fast, who knows which capabilities may be the best to use, hopefully safely, around health data and other data? You can suddenly find yourself locked into a supplier that, because of its relationships with other customers, is not able to give you necessarily the best solution that you might want. I do not put all these interactions down to a singular motivation or malign motivation, but simply to not thinking through the consequences of putting all your eggs in one basket, and not understanding that that particular sort of provider is going to have influence upon it from its other work, its other customers and its own agendas.

[00:18:51] Dominic Bowen: Phil, I will just take the opportunity to remind our listeners that if you prefer to watch your podcasts, The International Risk Podcast is always available on YouTube. Please go to YouTube and search for The International Risk Podcast, and you can find all our content there. Please remember to subscribe and like; it is really important for our long-term success.

[00:19:09] Dominic Bowen: Phil, NHS England has been quite clear in its statements that Palantir cannot commercialise, market, train AI on, or derive products from NHS data, and that it can only process data in accordance with NHS instructions. That seems good and reasonable. So why are you and other campaigners still concerned about this relationship? The risks you have talked about sound fair and reasonable. But NHS England has countered that by saying it cannot be commercialised, marketed, trained on, or used to derive products from it. So why should we still be concerned?

[00:19:42] Phil Booth: NHS England is simply wrong. I can give two examples: Cancer 360 and OPTICA. Palantir has already developed software which it can now sell to others, sitting on Foundry in another country, to manage their cancer care or patient discharge processes. That is something that has been developed by looking at the pipelines and the flows of NHS patients’ data through NHS processes. Unfortunately, this is really corrosive of trust. The NHS, NHS England and the government are constantly basically blowing a smokescreen for what they have actually done. To make a statement that they cannot create products, when they have and we can point to those products, is really a very bad position for a national body or a government to take. It is this level of dishonesty that really does not help in the ongoing debates around what should be done, and how things should be done, with data.

[00:20:40] Dominic Bowen: The Mayor of London, Sadiq Khan, just in the last few weeks blocked a proposed £50 million contract between the Metropolitan Police and Palantir, citing concerns around procurement practices, value for money and reputational risk. So it does make you ask: even if you ignored everything that you have been talking about for the last 20-odd minutes, if the police contract has been blocked by the Mayor of London because of reputational risk and concerns around procurement, why is the National Health Service allowed to use it? What did you hear when you heard Sadiq Khan say that, and what were some of the thoughts that went through your mind when you heard his statements?

[00:21:20] Phil Booth: I was obviously glad that the governance arrangements around this work did operate. In this particular situation, the Met Police are under a particular body, I think it is called MOPAC, and the Mayor’s Office and the Home Office. So there is, if you like, shared governance of decisions like this. Clearly, with all of the things that people are now learning about Palantir, and the concerns that were already there, this is a democratic process working. While the Met Police, and maybe the Home Office, would be perfectly happy with Palantir, clearly the public are not, and their democratic representatives are engaging in that debate. I note that in Germany, I think the BfV or so has decided to go with a French company, ChapsVision, instead of Palantir. So even the intelligence services are beginning to question whether or not they want to be using Palantir.

[00:22:14] Phil Booth: Do not forget, obviously, that Palantir started out post-9/11 as essentially a joint venture. Thiel and the CIA’s investment arm, In-Q-Tel, developed the original software, which I was actually given a demo of, as part of civil society in the UK, when Palantir started to sell its software outside of the US and to customers outside of the intelligence and security community. They came with this sort of promise: hey, we are civil-liberties friendly. They showed us the software. This was back in the early 2010s. They showed us the software and gave us a demo. It looked like very competent software to me, but there were obvious problems. The audit trail could simply be switched off. They would do whatever their customer asked them to do, which would not necessarily be consistent with the needs or rights of individual citizens. You have got to choose who your customers are. If your customer base includes some pretty nasty autocrats and dictators, then that raises questions about why you are selling to, or why you should be bought by, people in democracies.

[00:23:18] Phil Booth: These problems have been around for a good long while. To give them their due back then, as a consequence of maybe not just that meeting but meetings like those, Palantir itself instituted one of these internal ethics boards, I guess, to consider these sorts of issues. But as we have seen progressively over time, with a very controversial chief executive like Alex Karp making all sorts of inflammatory statements, they have not done themselves any favours. So I think it is good. I was glad to see that there is some democratic pushback. We are not just going to have these contracts signed through and waved through just because the Home Office or the police want something. That is not the only criterion. We have policing by consent in this country, and that consent, obviously in a different form than opt-outs and things like that, needs to be preserved. That cultural consent needs to be preserved in a time when the tools that police forces are using now may not be arming every single policeman with a gun in the UK, but they are arming them with really, really very weaponised pieces of software.

[00:24:30] Dominic Bowen: Phil, just in the last 30 seconds of the episode, one question that we ask all guests who come on The International Risk Podcast is: when you look around the world, when you look at all the things happening across the globe today, what are the international risks that concern you the most?

[00:24:43] Phil Booth: I would say that government is increasingly trying to turn people into numbers and deal with them like numbers, not human beings. We have historic evidence of what happens when you turn people into numbers. Unfortunately, it ends in ovens. We are not there yet. But the mass technologies that are being applied are starting to move us in that direction.

[00:25:08] Dominic Bowen: Thanks very much for explaining that, Phil. And thank you very much for taking the time to come on The International Risk Podcast today.

[00:25:13] Phil Booth: Thank you, Dominic.

[00:25:14] Dominic Bowen: Well, that was an interesting conversation with Phil Booth. I really appreciated hearing his thoughts on Palantir, on the NHS, patient confidentiality and the wider question that is relevant across all countries about who controls data in modern democracies. Today’s podcast was produced and coordinated by Edward Penrose. I am Dominic Bowen. Thanks very much for listening. We will speak again in the next couple of days.

[00:25:35] Outro: Thank you for listening to this episode of The International Risk Podcast. For more episodes and articles, visit The International Risk Podcast. Follow us on LinkedIn, Bluesky and Instagram for the latest updates and to ask your questions to our host, Dominic Bowen. See you next time.