“When states transfer more power to private entities, the public loses. The government becomes a body designed to service the needs and interests of profit-seeking private entities rather than its citizens. Worse, citizens lose much of their power to hold governments accountable: the government abdicates and transfers responsibility to private actors against which citizens have fewer rights.”[1]
This piece examines the creeping infrastructure of AI-powered surveillance in the UK and its implications for the Prevent strategy’s pre-crime framework.
Digital ID, Palantir and Oracle
In January of this year, the UK government dropped its plans for a mandatory digital ID. This came after almost 3 million signatures on a parliamentary petition opposing digital IDs.[2]
Several corporations were in talks of implementing the digital ID apparatus. Palantir Technologies, named after the all-seeing stone in The Lord of the Rings,[3] dropped out of the bid back in October, citing a lack of public support for the policy.[4]
Rumours of Oracle, described as a strategic partner of Palantir on its own website, receiving a contract for a digital ID app came as a result of the Tony Blair Institute backing the policy and the Blair family’s connections to Oracle.[5]
Both these tech companies have been at the forefront of surveillance concerns. The co-founder and Chairman of Oracle, Larry Ellison has stated that he envisions a state where citizens and police will be on their “best behaviour” because they will be constantly monitored.[6]
Palantir has been accused of many human rights violations, including Amnesty International’s findings that it was “targeting parents and caregivers of unaccompanied migrant children”.[7]
Although plans for a mandatory digital ID have been dropped, a voluntary digital ID will completely change the social and economic landscape of the UK.
It also sheds some light on the tech companies bidding on the rights to surveil citizens. These are the same companies whose technologies are already shaping counter-terrorism and pre-crime frameworks, including Prevent.
How is this related to Prevent?
With the advent of AI, it is important to consider how this will affect the UK’s Prevent strategy, one that is already couched in the pre-crime arena. Many of these tech companies have already been involved in counter-terrorism and predictive policing policies all over the globe.
Prevent already operates in the pre-crime space: referrals are made not on the basis of any criminal act, but on perceived risk or ideology. The introduction of AI into this framework would automate and accelerate that logic, potentially at population scale.
Criminal profiling as a tool traces its origins to the 20th century, especially the 1970s with the emergence of the FBI’s Behavioural Science Unit.
This involved making inferences about perpetrators of crimes by analysing behavioural and psychological traits. At the time, this was a breakthrough on understanding criminal behaviour.[8]
Data gathering today however, especially in the age of the internet, can be an overwhelming task. Information is seemingly endless and disparate. Gone are the days of men in headphones analysing hours of crackly wiretap audio in order to track down criminals. Today, with the capability of AI, everything becomes ‘searchable’.
“The vast amount of digital information now generated by the average individual means that more of this routine activity could be understood through analysis. Sources include communications metadata and internet connection records, but also extend to location and activity tracking, purchases and social media activity.”[9]
Every word in every conversation can be retrieved, every digital footprint cast in a net and filtered for relevance, and that too in a matter of seconds. In fact, now AI has the capability of analysing habits not just to summarise and provide insights, but to predict.
The History of Predictive Policing
AI has become the preferred tool for criminal profiling. Although optimists offer that the use of AI overcomes the limitations of intuition-based profiling, making findings more objective, AI comes with its own limitations.[10]
Predictive technologies used in the ‘war on terror’ post-911, were described as:
“‘preemptive lines of sight … visualizing unknown futures’, the ‘preemptive visualization of an unknown person,’ and ‘preemptive decision making’.”[11]
For example, in 2007, the US National Security Agency (NSA) used a system called SKYNET in Pakistan to predict involvement in terrorism. This was done through AI analysis of the metadata from 55 million domestic Pakistani mobile phone users. However, it faced issues of false positivity, wrongfully identifying 15,000 individuals as suspects.[12]
Amongst all these attempts, Palantir has come out as the preferred AI apparatus for the likes of:
JPMorgan, the World Food Programme (WFP) (the food assistance branch of the UN), NYPD,[14] the FBI, ICE, NYPD, Disease Control and Prevention (CDC), the Food and Drug Administration (FDA), NHS and the UK’s Royal Navy.[15]
From May 2014 to April 2015, the London Met trialled Palantir’s Predictive Crime Mapping products. Since then it is unknown whether the London Met has contracts with Palantir however over the years Palantir has kept a connection with the Met by attending or sponsoring receptions hosted by various contacts of the London Met. This included a sponsor of £3,000 in drinks for a promotional stand at a counter terrorism border conference in 2019.[16]
What is Palantir?
Palantir has been continually referred to as the, ‘most secretive and understudied surveillance firms globally’.[17] The name ‘Palantir ‘itself comes from “the crystal ball-like seeing stones from The Lord of the Rings, used for communication and to peer into the past and the future”.[18]
It was founded in 2003 by Peter Thiel, Nathan Gettings, Joe Lonsdale, Stephen Cohen, and Alex Karp.[19] The funding for the company came from In-Q-Tel, the CIA’s venture capital arm and remained somewhat of a secret till 2020, when it debuted on the New York Stock Exchange (NYSE).[20]
Many argue that Palantir has thoroughly embedded throughout the UK working with the Cabinet Office, the police, and the Ministry of Defence.[21]
The CEO of Palantir, Alex Karp stated in 2020 that the tools they develop are used to “kill people”.[22]
Gotham, Foundry and Apollo
Palantir offers 3 platforms as part of its surveillance eco system: Gotham, Foundry and Apollo.[23]
Gotham, formerly known as ‘Palantir Government’,[24] is the primary interface which allows the client to navigate through data. A now removed description from Palantir’s website stated that it provided a “full-text analysis for users to derive meaning from their organization’s data asset”.[25] It organises and centralises otherwise disparate and disconnected information into, “a single, coherent data asset”.[26]
Foundry is described as the more corporate-focused version of Gotham.[27] It is where users can “experiment and test preemptive decisions based on their data in a safe cloud environment to partition separate programs”.[28]
Apollo is described as, “the continuous delivery system that serves Foundry and Gotham”.[29]
However many users of Gotham and Foundry speak of them simply as the Palantir platform and in an interview with Robert Fink, a designer of Foundry, he stated that all products existed within the same he same unified database backend called AtlasDB.[30]
“The main function of both platforms is to collect and combine data stored and siloed in heterogeneous forms and from different databases into one single interface, thus allowing the client organisation to ‘know what they know’”.[31]
Metadata and Dynamic Ontology
There are two main functions of Palantir: producing ‘knowledge’ and ‘predicting’ events.[32]
This often involves leveraging metadata. Metadata is described as ‘data about data’ and “allow[s] people and machines to add context to primary data, including timestamps, geolocation, and authorship”.[33]
It is this labelling of the metadata which integrates previously separate datasets and centralises them in a manner that is browsable. This is something Palantir describes as ‘dynamic ontology’.[34]
The NHS and MoD
Palantir has entrenched itself into the UK, with two of its most major contracts, one with the NHS and the other with the MoD (Ministry of Defence) involving creating a specific platform for the NHS.
The NHS used Foundry as their front end platform for COVID-19 databases.[35] This was a 1 month contract in 2020 for a total of £1.[36]
This reinforces their official strategy:
“In the first phase, we typically acquire new opportunities with minimal risk to our customers through short-term pilot deployments of our software platforms at no or low cost to them. We believe in proving the value of our platforms to our customers. During these short-term pilots, we operate the accounts at a loss. We believe that our investments during this phase will drive future revenue growth.”[37]
This seems to have worked because after the 1 month contract in June, it was extended till October for £1 million.[38]
The type of data Palantir was contracted to manage included: personal contact details; work contact and employment details; race/ethnicity info; political affiliations; criminal history; and physical/mental health condition.[39]
In 2023, Palantir won a £330m contract to create a new data management system for NHS England, known as the Federated Data Platform (FDP).[40]
Ironically, the FDP is being developed as Palantir has been aiding in Israel’s ‘war on hospitals’. In 2024, Palantir signed a deal with the IOF. IOF operations have resulted in the systematic destruction of Gaza’s healthcare system and the detaining, torturing and murder of hundreds of healthcare workers.[41]
Similarly, in September 2020, in response to an FOI request, the MoD admitted to using Foundry and having a contract with Palantir.[42]
As of December 2019, it was reported that the MoD had at least £28m worth of deals with Palantir. One of the reported uses of the software was to slow down the rate of staff voluntarily leaving the Navy.[43]
One other product that Palantir offers is the Defence Intelligence Platform which the MoD may be reportedly testing. The software, “enables defence intelligence agencies to integrate disparate data sources into unified investigations for counterterrorism and intelligence workflows.”[44]
In the event of cancelling a contract with Palantir, Palantir may refuse to provide the analyses of their data in a form that can be used with the next system, as was the case with the NYPD.[45] Or they may chose to litigate as they did in 2016, winning a lawsuit against the Pentagon itself for a $206 million army contract.[46]
This effectively makes the NHS and the MoD perpetually reliant on Palantir’s platform once in use.
Why does this matter?
As a result of Edward Snowden’s revelations, it was revealed that Palantir was linked to surveillance operations employed by the US National Security Agency (NSA).[47]
In 2018, there were claims that the New Orleans Police Department (NOPD) used Palantir to secretly predictively police the population. It was used to identify individuals or places considered to be at high risk of crime.
A former law enforcement official stated: “They’re creating a target list, but we’re not going after Al Qaeda in Syria”.[48] ICE has used Palantir to “investigate potential targets and compile evidence against them”.[49]
The AI capabilities of Palantir, alongside its wide reach from the NHS to the MoD to the Home Office, creates a centralisation of information on a scale never seen before, and that too without consent of the public.
Proponents of AI argue that this centralisation is optimal for counter-terrorism efforts. It reduces the “human costs of war”; it can identify terror suspects in real time through facial recognition and even predict terror attacks.
Moving even further into the pre-crime sphere, some suggest that AI may, “support de-radicalisation programs by identifying at-risk individuals and analysing their online behaviour”.[50]
This would result in a completely new iteration of the PREVENT strategy, one with even more devastating human rights implications and constant mass surveillance.
Critics of the use of AI in a de-radicalisation process outline a number of obstacles, many of which we have previously identified in regard to the current PREVENT strategy itself:
- There is no real understanding of radicalisation.
“Radicalisation is a highly complex and individualised process and although researchers have identified certain elements reoccurring in radicalisation processes, there is not enough information to train an algorithm.”[51]
Nonetheless the current iteration of the Prevent strategy considers ideology as what underpins radicalisation. Furthermore, it particularly focuses on ‘Islamist’ ideology. By incorporating AI into the de-radicalisation process, it would grant a veneer of objectivity to ideological frameworks – and even then, in a fundamentally subjective and broken way.
This is exceedingly a cause for concern, given that the co-founder of Palantir, Peter Thiel has previously postulated about Islam as a ‘civilisational enemy’ to the West, borrowing from the likes of Carl Schmitt, the Nazi Jurist.[61]
Couple this with a CEO who, in a 2021 video that circulated widely publicly fantasises about using drones to spray ‘light fentanyl-laced urine’ on analysts critical of Palantir, and question marks must be raised about whether the use of AI will truly be of use in de-radicalisation or if it will aid in criminalising dissent in an increasingly Orwellian UK state.
- Current predictive policing systems that incorporate AI contain algorithmic prejudice and discrimination.[52] The US police department have been accused of using Palantir’s facial recognition technologies to racially profile individuals in order to predict crime.[53]
- Such an incorporation would be an affront to the civil liberties of citizens.
“As a thought experiment: what would be needed to predict individual behaviour? An algorithm predicting individual behaviour would need more differentiated data than the available group‑based information. Namely, it would require non‑anonymised data on the behaviour of individuals, the more the merrier, to ensure predictions are reliable. This would require a surveillance of individual behaviour in an unprecedented way: live‑coverage of an entire society would be necessary. This is not compatible with existing privacy rights. Nor is it desirable in a free society, for ethical and moral reasons. It would have repercussions on fundamental rights, such as freedom of speech, of the press, of association, of telecommunications secrecy and so on. Essentially, it would create nothing less than a dystopia.”[54]
For these reasons, inter alia, Kent police ultimately discontinued the use of PredPol, an American software used to predict crime.[55]
For individuals referred to Prevent, the stakes of this expansion could not be higher. An algorithmically-generated referral carries the same life-altering consequences as any other: loss of employment, family disruption, stigmatisation, and the chilling effect on lawful religious and political expression.
Prevent Watch continues to document these harms. The integration of AI into this framework would only serve to further entrench and accelerate the problems associated with Prevent.
Conclusion
As far back as 2014, the Office of the United Nations High Commissioner for Human Rights (OHCHR) expressed concerns about digital surveillance:
“Examples of overt and covert digital surveillance in jurisdictions around the world have proliferated, with governmental mass surveillance emerging as a dangerous habit rather than an exceptional measure.”[56]
One study around Palantir’s patent topics and the corresponding themes, highlighted:
“Palantir’s characterization of ‘adversaries’ and its critique of Silicon Valley’s understanding of ‘justice’ and ‘how society should be organized’ are curious when read alongside the patents presented in this article, where technologies for data gathering, labeling, and modeling are described in contexts that include facial recognition, predictive policing, workplace surveillance, and social media mining.”[57]
AI is a tool, however, increasing centralisation of data, and its use to identify, predict and preempt crime is not only concerning in the context of civil liberties, but can have catastrophic implications in the event of over-reliance.
Israel’s genocide in Gaza is a perfect example of the implications of the use of AI in preventative, de-radicalisation and counter-terrorism efforts:
“As it began targeting Hamas, it [Israel] deployed an ensemble of tools to detect and anticipate militants, which had a high false-positive rate, contributing to the death toll in Gaza.”[58]
In other words, a reliance on AI has resulted in the murder of innocent civilians falsely identified as Hamas operatives in Gaza. Furthermore, the over-reliance on AI, some have argued, was one of the factors that contributed to the October 7th attack.
“In the years leading up to October 7, Unit 8200 – Israel’s signal intelligence (SIGINT) body – reorganized itself, moving away from relying on traditional human analysts in favor of AI/ML tools designed to predict terrorist and insurgent activity.”[59]
As a result, the AI tools failed to raise the alarms necessary to warn leadership of the impending attack “even as a human analyst detected the plot and raised alarms”.[60]
In conclusion, when using AI tools to identify terrorist threats, those tools in practice will produce false-positives AND false-negatives which are both concerning in their own right.
References
[1] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 27
[2] https://www.bbc.com/news/articles/c3385zrrx73o
[3] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 5
[4] https://complexdiscovery.com/palantir-opts-out-of-uk-digital-id-initiative-amid-privacy-concerns/
[5] https://consoc.org.uk/time-for-digital-id/
[6] https://fortune.com/2025/09/28/larry-ellison-ai-surveillance-oracle-tiktok-deal-social-media/
[7] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 334-335
[8] Taylor Smith, ‘Intuition Vs. Data: The Role of Human Judgement in Pre-AI Criminal profiling’ [2025] 1
[9] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 338
[10] Taylor Smith, ‘Intuition Vs. Data: The Role of Human Judgement in Pre-AI Criminal profiling’ [2025] 5
[11] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 338
[12] Kathleen McKendrick, ‘Artificial Intelligence Prediction and Counterterrorism’ [2019] 11
[14] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[15] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 336
[16] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 21
[17] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 334
[18] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[19] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[20] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[21] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 8
[22] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 8
[23] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 338
[24] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 9-10
[25] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 339
[26] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 9-8
[27] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 10
[28] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 339
[29] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 339
[30] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 12-13
[31] Vasileios-Spyridon Vlassis, ‘More than software vendors. The peculiar case of Palantir’s data integration platforms’ [2024] 6
[32] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 336
[33] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 336
[34] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 339
[35] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 14
[36] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 14
[37] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 15
[38] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 15
[39] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 15-16
[40] Rhiannon Mihranian Osborne, ‘NHS England must cancel its contract with Palantir’ [2024] 1
[41] Rhiannon Mihranian Osborne, ‘NHS England must cancel its contract with Palantir’ [2024] 1
[42] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 23
[43] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 22
[44] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 24
[45] Mallika Balakrishnan, Niyousha Bastani, Matt Mahmoudi, Quito Tsui, & Jacob Zionts, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ [2020] 16
[46] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[47] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 334
[48] Mustapha Jid, ‘Palantir’s Predictive Policing Technology: A Case of algorithmic Bias and Lack of Transparency’ [2023] 2
[49] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[50] Alice Martini, ‘Developing CTS 4-o: a CTS research agenda on AI, terrorism and counter-terrorism’ [2025] 541
[51] Marie Schroete ‘Artificial Inteligence and Countering Extremism: A Primer’ [2020] 22
[52] Mustapha Jid, ‘Palantir’s Predictive Policing Technology: A Case of algorithmic Bias and Lack of Transparency’ [2023] 2
[53] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 335
[54] Marie Schroete ‘Artificial Inteligence and Countering Extremism: A Primer’ [2020] 22
[55] Marie Schroete ‘Artificial Inteligence and Countering Extremism: A Primer’ [2020] 21
[56] Kathleen McKendrick, ‘Artificial Intelligence Prediction and Counterterrorism’ [2019] 12
[57] Andrew Iliadis & Amelia Acker, ‘The Seer and The Seen: A Survey of Topics Found in Palantir Patents’ [2022] 358
[58] Christopher Wall, ‘The Ghost in the Machine: Counterterrorism in the Age of Artificial Intelligence’ [2025] 16
[59] Christopher Wall, ‘The Ghost in the Machine: Counterterrorism in the Age of Artificial Intelligence’ [2025] 15
[60] Christopher Wall, ‘The Ghost in the Machine: Counterterrorism in the Age of Artificial Intelligence’ [2025] 16
[61] https://coolnessofhindblog.wordpress.com/2025/02/10/zio-technofascism-elon-musk/
Photo by Markus Spiske on Unsplash
