Category: Trump

Our data, their power: the growing reach of Palantir

Palantir is building a future in which our most intimate information becomes a tool of power. Who is watching the watchers?

The US data company has secured a £330m NHS contract and works with military and government agencies. The question is no longer simply what its software can do, but who should be trusted to control the information on which our lives depend.

Every time we visit a hospital, apply for a public service or interact with the state, we leave behind information about ourselves. Our medical histories, personal circumstances and everyday decisions form an increasingly detailed picture of who we are. The question is who gets to connect those fragments, analyse them and use what they reveal.

That question has become increasingly urgent as Palantir, the American data analytics company co-founded by billionaire Peter Thiel, expands its role in British public services. The company is at the centre of a growing controversy over a £330m contract with the NHS to provide its Federated Data Platform (FDP), software designed to help health services bring together information from different systems and make operational decisions.

In July 2026, MPs on the health and social care select committee urged the government to prepare an alternative, citing serious public mistrust, questions about the platform’s effectiveness and concerns over patient privacy. A separate parliamentary committee had already called for the government to consider ending the arrangement.

The government is reviewing the contract ahead of a break clause in February 2027. The decision will have implications far beyond the procurement of a piece of software. It will help determine how much control the public retains over the information entrusted to its institutions

A company with extraordinary access

Founded in 2003, Palantir grew out of the world of intelligence and national security. Its early funding included investment from In-Q-Tel, the investment organisation associated with the CIA. Its business has since expanded across government, healthcare, defence and commercial industries.

Thiel, a prominent supporter of Donald Trump, is one of the company’s co-founders. Trump himself did not found Palantir, but the company’s political associations have become part of a wider debate about the growing influence of powerful technology firms over public institutions.

Palantir’s software is designed to integrate data held in separate systems, allowing authorised users to identify connections and patterns that might otherwise remain hidden. That can be useful in complex organisations, where information is often fragmented and decisions need to be made quickly. But the same capability raises difficult questions. When information collected for one purpose can be analysed alongside information collected for another, the picture that emerges can be far more revealing than any individual record.

The concern is not simply that a company might possess a database containing everyone’s personal information. Palantir’s NHS platform is intended to connect existing systems rather than copy all their data into one central repository. The company says its software processes information according to the instructions and controls of the organisations using it. Nevertheless, the ability to connect, organise and analyse sensitive information can itself confer considerable power. A system does not need to own every record to influence how those records are interpreted and acted upon.

The NHS holds some of the most sensitive information imaginable: diagnoses, treatments, prescriptions, mental health records and details of deeply personal circumstances. Used responsibly, this information can improve care, reduce delays and help hospitals plan their services. Better coordination between departments can mean fewer missed appointments, more efficient use of operating theatres and patients receiving treatment sooner. Palantir and NHS officials have pointed to claimed improvements associated with the platform, including additional operations and reduced discharge delays. These benefits deserve serious consideration, just as claims about performance should be independently scrutinised.

But efficiency cannot be the only measure of success. Patients share information with doctors because they expect it to be used to treat them, not because they have knowingly agreed to every future use of their data by every technology provider involved in running the health service. There is a fundamental difference between allowing a hospital to access information needed for treatment and accepting an expanding infrastructure that makes it easier to combine and analyse information across an entire health system.vThat difference matters even when access is legally authorised and safeguards are in place. The public must be able to understand what information is being processed, who can access it, what decisions it informs and what happens if the system is misused.

In July, up to 117 NHS data and technology workers called for the contract to be abandoned, warning that inadequate privacy protections and declining public trust could undermine the completeness of health data. Their concern points to a serious practical risk: if people become reluctant to share information with the NHS because they do not trust the systems handling it, the consequences could affect healthcare itself. Palantir’s work extends beyond the NHS. The company has contracts and relationships involving defence, law enforcement and government agencies, including work associated with the US immigration authorities and the Israeli military. These uses are not identical, and the existence of a contract does not establish that every application of the software is abusive. But they demonstrate why the company’s technology cannot be evaluated solely as a neutral administrative tool.

Software that helps a hospital coordinate care may also help a government agency connect information, identify individuals and inform enforcement decisions. The consequences depend on the data involved, the authority using the system, the purpose of the analysis and the safeguards governing the result. That is precisely why democratic scrutiny matters.

When a company supplies technology to public institutions, its influence can extend beyond the terms of an individual contract. Public bodies may become dependent on its expertise, technical infrastructure and proprietary systems. Replacing a platform can then become expensive and complicated, even if public confidence deteriorates. This creates a question that should concern every government: how do we prevent essential public services from becoming dependent on private companies whose systems are difficult to inspect, challenge or replace?

Who controls the information?

Palantir says its software operates under the control of the organisations that use it, with data processed according to their instructions. That assurance is important. It is also not the end of the debate. Formal ownership and legal control do not automatically answer every question about practical power. Who determines which connections the software makes? Who can audit its operation? Can public authorities independently verify how information is handled? What happens when a contract ends? And can the public obtain meaningful explanations when an automated or data-driven process affects them?

These are questions for Parliament, regulators, the NHS and the company itself. They also require clarity about the limits of the technology. Data analysis can identify patterns, but patterns are not always explanations. Information may be incomplete, inconsistent or shaped by historical inequalities. Decisions made using flawed information can reproduce those problems at scale.

The more consequential the decision, the greater the need for accountability, independent scrutiny and a meaningful route to challenge mistakes.The debate over Palantir is not an argument against technology, nor is it a claim that every private contractor presents the same risks. Digital tools can improve public services, and refusing to modernise systems carries costs of its own. But public institutions have obligations that private companies do not share in quite the same way. The NHS exists to serve patients, not to maximise shareholder returns. Its legitimacy depends not only on the quality of the care it provides, but also on the confidence people have that their information will be handled responsibly.

If a technology provider can offer genuine improvements, those benefits should be demonstrated with transparent evidence. If equivalent alternatives exist, they should be assessed fairly, including options developed in-house or supplied by UK-based organisations. Cost, performance, security, long-term dependence and public trust should all form part of the calculation. The £330m contract should therefore be judged on more than whether the software works. It should be judged on whether the NHS can retain meaningful control over its digital infrastructure, whether patients can trust the protections around their information and whether the public interest is adequately safeguarded.

Palantir did not invent the collection of personal information, and it does not literally take every piece of information about every person. The deeper concern is the expanding capacity of powerful systems to bring separate records together and turn them into actionable intelligence.That capacity is increasingly valuable to governments, hospitals, police forces and military organisations. It can help solve real problems. It can also concentrate power in ways that ordinary citizens struggle to see, understand or contest. The British government now faces a choice about how it wants this technology to operate within the public realm. The question is not whether data can be useful. It plainly can. The question is what limits should govern its use, who enforces those limits and what happens when public confidence is lost. Our information tells the story of our lives. It should not become a resource over which citizens have little visibility or influence simply because the technology to analyse it has become available.

A modern public service needs effective digital systems. It also needs public consent, independent oversight and the ability to change course.

Without those protections, the promise of efficiency risks overshadowing a more fundamental principle: the information people entrust to the state must remain subject to democratic accountability.

The Midterms Are Beginning to Look Like a Referendum on Trump

There is a particular point in an American election campaign when what once looked like a collection of interesting polls begins to resemble a genuine political trend. We may now be approaching that point in the 2026 midterm elections. The election is still a month away, and American politics is famously capable of producing extraordinary reversals in a remarkably short period of time, but the evidence accumulating across the country is becoming increasingly difficult for Republicans to dismiss. Democrats are ahead on the national congressional ballot, their voters appear considerably more engaged, President Donald Trump’s approval rating has fallen to deeply uncomfortable levels, and Republican candidates in competitive races are increasingly behaving as though the president’s name is something they would rather keep at arm’s length.

At this stage, I would be genuinely surprised if Democrats did not win the House of Representatives. The Senate is a more complicated proposition because the electoral map gives Republicans structural advantages that do not disappear simply because the national political environment has turned against them. Yet even there, the closer we get to November 3, the harder it becomes to dismiss the possibility of a Democratic majority. What looked like an ambitious Democratic objective earlier in the campaign is beginning to look like a perfectly realistic outcome. And there is one person who sits at the centre of almost all of it: Donald Trump.

The latest polling is not merely showing Democrats with a small advantage. It is showing a pattern. Recent national congressional polling has Democrats ahead of Republicans by a significant margin, with several surveys showing advantages that would be more than sufficient to put a large number of Republican-held House seats into play. Pew Research Center has also found Democrats with an advantage on the congressional ballot, while Democratic voters appear more engaged with the coming election than Republicans. That enthusiasm gap matters enormously in a midterm election because turnout is lower than during a presidential election and relatively small differences in motivation can have enormous consequences in individual congressional districts and states. A party that enters a midterm election with an enthusiastic base while the president occupying the White House is unpopular is already starting with a significant advantage.

The House of Representatives is where the Democratic case is strongest. The Republican majority is narrow, and narrow majorities are notoriously difficult to defend in midterm elections when the party occupying the White House is unpopular. A president’s party has historically tended to lose seats during midterms, and when the president is particularly unpopular, the losses can become much larger. Republicans are therefore not simply trying to defend a comfortable majority against a reasonably strong Democratic opposition; they are trying to defend a narrow majority while their president is unpopular, while Democrats are performing strongly on the national congressional ballot and while a number of previously safe Republican districts are becoming competitive.

This is why I think Democrats should be looking beyond the question of whether they can scrape together enough seats to take control of the House. The political environment increasingly suggests that they should be thinking about whether they can win convincingly. A narrow victory would be enough to change the balance of power in Washington, but a large Democratic victory would tell a much more important political story. It would demonstrate that the electorate had not simply moved against a handful of vulnerable Republican representatives but had turned decisively against the broader political direction of the Republican Party and the Trump administration.

The significance of that distinction should not be underestimated. If Democrats begin winning Republican-held districts that would normally be considered safely conservative, the election starts to look less like an ordinary midterm contest and more like a national wave. Republican candidates are then forced to spend money defending seats they never expected to have to defend, while Democrats can use their resources to attack increasingly ambitious targets. That creates a vicious cycle for the party defending the majority because every additional seat that becomes competitive forces Republicans to divide their resources further.

The Senate is a much more complicated proposition. The electoral map is more favourable to Republicans than the House map, and Democrats have fewer obvious opportunities. Several Republican-held Senate seats are in states where Donald Trump remains capable of winning or where the underlying political structure remains favourable to Republicans. It would therefore be premature to describe a Democratic Senate victory as inevitable.

But there is a significant difference between saying that a Democratic Senate victory is not inevitable and saying that it is unlikely. Those are no longer the same thing. Several current forecasting models are giving Democrats a genuine chance of taking control of the chamber, and that possibility becomes more credible if the national political environment continues moving in the same direction. If Democrats maintain a substantial advantage on the generic congressional ballot, if their voters remain more enthusiastic about participating, and if Trump’s approval rating remains deeply underwater, the Republican Party may find that its structural advantage in the Senate map is simply not enough to protect every vulnerable seat.

That is why the Senate could ultimately become the biggest surprise of election night. Earlier in the campaign, Democratic control of the Senate would have looked like a difficult objective requiring almost everything to go right. As we get closer to November, however, it is becoming increasingly difficult to dismiss. If the current political environment persists, I would be genuinely surprised if Democrats did not at least come extremely close to taking the chamber, and I would be even more surprised if Republicans managed to emerge from the election having comfortably retained both chambers of Congress.

The most interesting aspect of this election may be the increasingly obvious problem Republicans have with their own president. Donald Trump remains an extraordinarily powerful political figure, and there is no question that his support can still be enormously valuable to Republican candidates. His endorsement carries considerable weight with the Republican base, his rallies can generate enormous enthusiasm and his ability to dominate the political conversation remains unmatched. But the very qualities that make Trump such an effective political force within the Republican Party can become liabilities when candidates need to appeal beyond the Republican base.

This is the strange paradox of Trumpism. Trump is exceptionally good at energising people who already love him, but he is also exceptionally good at energising people who absolutely do not. In a presidential primary, the first group can be enough to determine the outcome. In a general election, particularly in competitive suburban congressional districts and closely contested Senate states, the second group can be just as important.

Recent polling has put Trump’s approval rating at around the low 30s, which is an extraordinarily difficult position for a president’s party to be in ahead of a midterm election. The problem is not simply that some voters dislike him. It is that Trump’s political identity has become so closely intertwined with the Republican Party that Republican candidates cannot easily separate themselves from him even when doing so might be electorally advantageous.

That leaves Republican candidates with a remarkably difficult choice. If they embrace Trump completely, they risk inheriting his unpopularity among independent and moderate voters. If they distance themselves from him, they risk alienating the Republican voters who remain intensely loyal to him. There is no obvious way out of the dilemma because Trump has spent years transforming the Republican Party into a political movement centred around his personality, his political instincts and his particular view of American politics.

Recent reporting has highlighted just how significant this problem has become. Republican candidates in competitive congressional races are increasingly distancing themselves from Trump on particular issues, even while remaining members of a party that is still fundamentally shaped by his presidency. That tells you something important. These candidates may not want to admit publicly that Trump is hurting them, but their campaign strategies increasingly suggest that they understand the risk.

There is something almost absurd about the position in which some Republican candidates now find themselves. Trump wants the midterms to be a referendum on his presidency. He wants Republican candidates to embrace his agenda and defend his record. Yet candidates running in competitive districts increasingly want voters to judge them independently. They want to talk about local issues, their own records and the concerns of their constituents rather than spending every day defending the latest controversy surrounding the president.

The problem is that they cannot completely escape him. Trump is the president, the leader of their party and the central figure in Republican politics. Democrats can therefore continue asking a very simple question: if you vote Republican, aren’t you also voting to give Donald Trump more power?

That is a potentially devastating argument in a midterm election. Voters do not need to dislike their individual Republican representative in order to vote against them. They simply need to believe that Congress should provide a check on the president. If Democrats successfully turn the election into a referendum on whether Trump should have more or less power, Republican candidates could find themselves fighting a national battle that many of them would rather have avoided.

The more Trump makes the election about himself, the harder it becomes for Republican candidates to make the case that they should be judged independently. This is perhaps the greatest political irony of the moment. Trump spent years arguing that the Republican Party needed to become more unified around a clear political identity, and he succeeded spectacularly. But that success means that when voters become dissatisfied with Trump, Republican candidates have very little room to escape the consequences.

The other major issue is the economy. Elections are rarely decided entirely by ideology, particularly in midterms, and voters tend to become much more unforgiving when they feel that their everyday financial circumstances are deteriorating. Inflation, food prices, housing costs, energy prices and wages are considerably more tangible to voters than most of the ideological arguments that dominate political coverage.

This gives Democrats an obvious opportunity. They can argue that Americans should judge the Trump administration according to a simple question: are you better off than you were before? If voters answer no, Democrats do not necessarily need to persuade them that every Democratic policy is superior. They simply need to persuade them that Republicans deserve to be punished for the economic situation.

That is a much easier argument to make in a midterm election than trying to persuade voters to embrace an entire Democratic ideological programme. The opposition party can campaign as a check on the administration rather than presenting itself as an alternative government-in-waiting, and if Trump’s approval remains around the low 30s, that strategy could be remarkably effective.

This is also where Trump’s political style could become particularly damaging. His presidency is rarely quiet. Every day brings another controversy, another confrontation, another argument and another political battle. That style is extraordinarily effective at maintaining attention, but constant political conflict can eventually become exhausting to voters who are not particularly interested in politics. A voter can agree with some of Trump’s policies while simultaneously deciding that the country simply needs less drama.

That exhaustion should not be underestimated.

It would nevertheless be a mistake to assume that the election is already decided. American politics has a remarkable capacity for surprise, and there is still plenty of time for Republicans to change the trajectory of the race. Trump remains capable of dominating the news cycle, energising his supporters and turning an obscure congressional contest into a national culture-war battle. His supporters remain extraordinarily loyal, and Republican voters who are currently less enthusiastic could become much more motivated as election day approaches.

The biggest danger for Democrats is therefore complacency. A favourable national environment does not automatically translate into a Democratic Senate majority, and even the House should not be treated as a certainty. The Senate map remains difficult, Republican candidates can outperform Trump, and voters who dislike the president can still vote for Republican candidates.

Democrats also have their own problems. Republicans retain advantages on several important issues, particularly immigration and crime, and Democrats cannot simply assume that every voter who is unhappy with Trump will automatically vote Democratic. They need to give those voters a reason to support them rather than simply a reason to oppose Republicans.

That distinction will ultimately determine whether this becomes an ordinary Democratic victory or a genuine political wave.

Republicans will argue that Democrats are too far to the left, too weak on immigration, too willing to obstruct the president and too committed to policies that they believe have failed. Democrats will argue that Republicans have given Trump too much power and that the only way to restore a meaningful check on the administration is to hand Congress back to the opposition.

The Democrats therefore have a potentially powerful argument that goes beyond individual races. They can tell voters that they do not have to decide whether they love or hate the Democratic Party. They simply have to decide whether they want Donald Trump to have an unchecked Republican Congress behind him for the remainder of his presidency.

If enough voters answer no, Democrats could have an extraordinary night.

And that is why the closer we get to November, the more surprised I would be if Democrats did not convincingly win the House. The national environment is simply becoming too difficult for Republicans to overcome, particularly when they are defending a narrow majority while their president remains deeply unpopular. The Senate is obviously harder, and there are still several races where Republicans have significant structural advantages, but the idea of Democrats taking the chamber is no longer some distant fantasy. It is increasingly a plausible outcome.

There is an extraordinary irony at the centre of all of this. Donald Trump has spent more than a decade transforming the Republican Party into a movement that revolves around him. He defeated the traditional Republican establishment, reshaped the party’s electoral coalition, changed its rhetoric and made his personal political brand inseparable from Republican identity.

That transformation produced enormous electoral benefits. But it also created an enormous vulnerability.

A political party that is closely identified with one person becomes extraordinarily powerful when that person is popular, but extraordinarily vulnerable when that person becomes unpopular. Republican candidates cannot simply pretend that Trump does not exist because their own voters are too closely connected to him. At the same time, they cannot always afford to embrace him because independent voters and moderate Republicans may be increasingly uncomfortable with the president.

This is the fundamental problem with having Trump on your side. His endorsement can win you a primary. His supporters can provide an enormous electoral base. His political machine can generate money, attention and enthusiasm. But when the general electorate turns against him, all of those advantages can become liabilities. The very voters who make Trump such a formidable force within the Republican Party can become a ceiling on the party’s ability to expand beyond its existing coalition. And if the current polling is correct, that ceiling may be becoming increasingly important.

There is still time for this election to change. There will be unexpected events, damaging advertisements, political scandals, candidate mistakes and probably several moments that completely dominate the news cycle for a few days. American elections are not predictable enough to declare the result before voters have actually cast their ballots.

But elections are ultimately about probabilities rather than certainties, and the probabilities are becoming increasingly favourable to Democrats. The House should be theirs. If the current environment continues, I would be surprised if they did not win it convincingly rather than simply scraping across the line.

The Senate is harder to predict, but the possibility of a Democratic majority is becoming increasingly difficult to dismiss. If Democratic enthusiasm remains high, if Republicans continue distancing themselves from Trump in competitive races, and if Trump’s approval remains around the low 30s, the structural advantages Republicans possess in the Senate may not be sufficient to protect every vulnerable seat.

The irony is that Trump may have spent years building the most personally loyal political movement in modern American history, only to discover that loyalty has a weakness. It can make a party extraordinarily powerful when its leader is popular, but it can make the entire party extraordinarily vulnerable when the electorate turns against him. For Republicans, that may be the defining problem of the 2026 midterms. They have Donald Trump on their side.

And increasingly, that may not be the advantage they once thought it was.