The evolution of national security is linked to the evolution of technology. As technology advances, it becomes a tool for governments to optimize and automate aspects of defense so that humans are not required to perform mundane, mechanical tasks. This allows people to shift focus to building higher-level strategy that requires human judgment. The advent of AI, however, has brought the latter part of the equation, i.e. the human judgment requirement, into question. States are increasingly integrating AI into not just the technical aspects of defense, but also the strategic aspects. This extends to both traditional and non-traditional aspects of national security. Take for example the tech company Palantir. US Immigration and Customs Enforcement (ICE) gave Palantir a contract to build a surveillance platform and parse the surveillance data to identify immigrants for deportation under ICE’s draconian criteria – targeting “subjects of national security concern,” according to ICE’s official website. In another instance, US’s Project Maven, the military’s flagship AI initiative, is inserting AI “into the kill chain,” as New Yorker staffer Gideon Lewis-Kraus writes. The tech company Palantir is heavily involved in the logistics of this project. Palantir’s use of AI in both these cases has concerning implications for the future of America’s defense autonomy, ironically by increasing the influence of autonomous AI in national security strategy.
ICE and Palantir’s ImmigrantOS
The second Trump administration emboldened anti-immigration agencies such as ICE to aggressively detain and deport immigrants involved in activities ranging from abject criminality such as human trafficking to drugs smuggling. Apart from this, ICE targets those who fraudulently benefit from immigrant documentation as well. ICE benefits from minimal bureaucratic disturbance in the arbitrary manner by which they classify and take action upon immigrants. Syracuse University’s Transactional Records Access Clearinghouse (TRAC) collated statistics relating to ICE detainment and other operations. Approximately 70% of ICE’s 60,311 detainees are not even criminally convicted, with an additional 180,701 people being monitored under the ICE Alternatives to Detention (ATD) program, as of April 2026. The tyranny of ICE also extends to US citizens seeking to defend innocents from arbitrary action. ICE classifies a wide range of people as issues of national security. However, the crucial question of human security remains in a limbo. By protecting “national security,” ICE is making everyday citizens face insecurity.
ICE awarded a $30 million contract to Palantir last year in order to build a surveillance platform called “ImmigrantOS,” according to the Immigration Policy Tracking Project. This platform would help ICE in parsing the surveillance data and classifying individuals needing to be apprehended, tracking self-deportation, and making deportation practices more structured and efficient. It would function as a commercial ISR (Intelligence, Surveillance, and Reconnaissance) tool in bolstering national security, given that people falling under ICE’s detainment and deportation criteria are classified as threats to national security.
Stay ahead of the geopolitical week.
MD Briefing delivers expert analysis across five global fronts — the Indo-Pacific, energy, geoeconomics, European security, and the Middle East — every Monday morning. Free.
Mass surveillance of the type required for such a system to work is in gross violation of the right to privacy, under Article 12 of the Universal Declaration of Human Rights (UDHR). In addition to this, the operating system that such a platform would run on would allow for a peremptory machine evaluation of individuals requiring ICE persecution, which removes the human judgment aspect from the entire operation. Such a system would amplify the numbers of innocent individuals falsely arrested. While there appears to be limited public information about the actual dissemination of this tool, lawyer groups are working to actively combat such illegal profiling. In essence, while the very foundation of ICE and its methods is problematic, the usage of such a surveillance tool would exacerbate the purview of ICE’s human rights violations. It holds as a grave warning against using such a tool to compromise the safety of the common people.
The Maven Smart System
Project Maven is the 2017 US initiative to integrate AI into American military capabilities. Its primary industry partner was Google, up until the 2018 employee protest against Google’s involvement in the war machine. Palantir became involved as Project Maven’s industry partner in 2019 with a $161 million partnership in that year. According to CSIS’s June 2026 report, the project’s flagship technology, the Maven Smart System (MSS) is an AI initiative processing information from 179 data sources through ISR, as of 2024, and allowing for operationalisation through target acquisition. After flagging targets, the software also enables commanders to move beyond simply receiving the target data by evaluating the environment and judging for actual strikes. Corresponding ISR allows users to assess the effects of the strike. The Chief Digital and Artificial Intelligence Officer of the Department of War, Cameron Stanley, mentions how the system is used to “overcome differences in mass and scale” against “adversaries.”
Currently, the software is not fully autonomous, keeping the “human-in-the-loop” (HITL) aspect intact. However, reducing human involvement in parsing the target criteria has its own disadvantages, especially since the HITL aspect only applies to the very last level of decision-making. The target acquisition capacities of such a system are susceptible to AI errors and biases, leaving potential to mislead the human in the loop. In addition to this, the CSIS report states that the use of MSS in the Iran war guaranteed the strikes of over 1000 targets, “a tenfold increase” from prior to MSS. MSS increases strike capacities and by corollary, collateral damage as well. Fundamentally, the tool encourages the war machine and indiscriminate targeting – something that should be treated with caution to begin with.
Conclusion
The case studies of Palantir’s ImmigrantOS and MSS demonstrate detrimental potential, whether it’s the decreasing scope of privacy and other human rights, or indiscriminate strike capacities in times of war, resulting in mass collateral damage. With AI’s capacity for error and lack of human judgment, the propensity for pulling the trigger at the wrong moment is an omnipresent possibility. While the trigger is currently metaphorical, increasing automation and AI reliance suggests that this propensity may become an actuality. The destructive potential of such tools serves as a cautionary tale in itself.
