Why the NHS Rejected Palantir's Platform

On October seventeenth, the Guardian published freedom of information responses showing sixteen NHS trusts and health boards in England are not using Palantir's patient data platform. Some tried it and went back to their old systems.

Sixteen out of how many, though? Because England has hundreds of trusts.

That's fair, and it's not a majority revolt. But it lands three weeks before ministers decide whether to kill the contract. There's a break clause in February twenty twenty-seven, and the government has said it'll decide by December.

Okay. So what are the trusts actually saying? "We don't like Peter Thiel"?

Mostly not, actually. University Hospitals of North Midlands said it "does not appear to provide functionality over and above what we have already". Dorset HealthCare said it will not be adopting it. A Suffolk trust said its own systems supersede it.

So it's a shrug, not a protest.

In those cases, yes. Greater Manchester's care board is the blunt one. They called the Palantir system "duplicative" and "retrograde".

Retrograde is a strong word for software.

Greater Manchester built their own thing over nearly a decade. Wired covered them back in August. Their chief data officer, Matt Hennessey, said moving to Palantir would be a backwards step, and the board reckoned some of their functions are two to three years ahead.

And Palantir says what to that?

Palantir and the platform's supporters dispute it. And the Guardian's own reporting is mixed: some health bodies said the platform made managing waiting lists easier and improved surgery scheduling.

Right, so it does work somewhere. Quick bit of background for me, because I've lost the thread. When did this even start?

The Palantir-led consortium was chosen in autumn twenty twenty-three. Rollout began March twenty twenty-four. The contract's worth up to three hundred and sixty million pounds over five years, with extensions that could take it to four hundred and eighty million.

And the Guardian keeps saying three hundred and thirty million.

Different ways of counting the same deal, which tells you how murky this contract is. One more thing worth knowing: the FOI responses were gathered by Medact, a health campaign group that opposes Palantir's role in the NHS, and published on WhatDoTheyKnow.

So campaigners went looking for nos and found nos. That doesn't make the quotes fake, but it's not a neutral survey.

No, it isn't. Which is why the second thing matters more. On October twentieth, Foxglove — a tech justice non-profit — published a report on Palantir's NHS work. Their central claim is that the headline success numbers don't stand up.

Which numbers?

Ministers and Palantir people kept citing extra operations delivered thanks to the platform. It started at eighty thousand, then a hundred thousand, then a hundred and ten thousand.

It grew in the telling.

And Foxglove says the government has admitted those headline claims can't be stood up. Their phrase is "we cannot draw conclusions about cause and effect". Foxglove also says a significant proportion of trusts using the tools saw performance decline.

Hang on. If the government admitted that, why were the claims still being made?

Foxglove alleges ministers kept repeating them to Parliament afterwards, and that the claims were still on NHS England's website as of September. That's their account, and it's a serious allegation.

What's the lobbying bit? The report title mentions lobbying.

Two things, both Foxglove's claims. One, the government held private meetings with Palantir about how to address the company's poor public reputation. Two, that it offered to look again at the rules on confidential patient information, to remove what Palantir called obstacles.

That second one would bother me as a patient. Who decides what's an obstacle?

And people have voted with their records. The national data guardian, Nicola Byrne, wrote to the health innovation minister James Frith. She said opt-out rates have risen alongside media coverage of Palantir, though she's careful: the cause can't be definitely established.

She's the official watchdog, right? Not a campaigner.

Right. And her line is striking. She wrote that the level of public opposition is unusual for a public sector software supplier, and that reassurances about legal and technical safeguards haven't moved people.

So where does the political decision actually sit now?

Frith wrote to the Commons Science and Technology Committee on October nineteenth, and there was further correspondence published on the twenty-first. Two select committees — Health and Social Care, and Science and Technology — have already urged the government to use the break clause.

And Andy Burnham's government has to pick.

By December, on their own timetable. The break clause itself is February twenty twenty-seven. If they don't use it, the contract can run to twenty thirty-one.

What would change your read on it?

Whether the government corrects the record on those operation numbers. Foxglove's asked for that explicitly, and so far I haven't seen a response.

Six weeks to decide on a seven-year contract. Someone's having a bad autumn.