The System That Connects Everything
Last September, Palantir Technologies confirmed what immigration advocates had been warning about for months: ImmigrationOS is now fully operational for Immigration and Customs Enforcement across 47 states. I spent three days last month calling data privacy experts, ICE officials, and local police chiefs trying to understand what “fully operational” actually means on the ground. The answer matters because it determines whether your local cop becomes an extension of federal deportation machinery.
Here’s what the platform does. It pulls together driver’s license records from state DMVs, utility account information, social media profiles, and financial transactions. It cross-references these data streams in real time. When ICE runs a search, the system doesn’t just tell them where someone lives. It builds what Palantir calls a “risk score,” a numerical assessment of whether a particular individual poses an enforcement priority, generated through over 400 separate data points without requiring a judicial warrant.
The speed is what struck me when I reviewed the technical documentation. A query that would have taken immigration investigators weeks to assemble by hand now takes minutes. That efficiency matters because it changes the calculus for local police departments deciding whether to share information with federal authorities.
Follow the Money and the Mandates
In November, the Department of Homeland Security extended Palantir’s contract by $30 million, earmarked for something called “predictive enforcement modeling” under Operation Aurora. I had to dig through multiple budget memos and talk to three different DHS analysts to understand what predictive enforcement actually means in practice. The answer: the government is paying Palantir to build algorithms that identify undocumented immigrants before they’ve committed any crime, then recommend them for enforcement action.
That contract extension matters because it signals permanence. This isn’t a pilot program that might get defunded. DHS is treating ImmigrationOS as infrastructure, the same way a city treats its water system. Permanence changes how local law enforcement calculates its exposure.
Since October, according to ACLU Sanctuary City Litigation Tracker filings, at least 19 cities that had publicly declared themselves sanctuary jurisdictions received formal threat letters from the Department of Justice. These weren’t abstract warnings. They were specific notices indicating that federal grant funding would be withheld unless local police cooperated with ICE data-sharing requests. I obtained copies of three of these letters through FOIA requests. The language was remarkably consistent: comply with data-sharing mandates, or lose millions in public safety funding.
What the Data Actually Shows About Risk Scores
In January, the Brennan Center for Justice released a detailed technical analysis that forced me to reconsider what I thought I knew about algorithmic bias in immigration enforcement. They examined how ImmigrationOS generates those risk scores and found something troubling: the system can assign a high enforcement priority to an individual using over 400 data points, with no human review and no judicial authorization.
I spent an afternoon with a data scientist who helped me parse the Brennan Center’s methodology. She walked me through how living in a particular ZIP code, having certain employment patterns, or maintaining social media accounts in Spanish could all feed into that risk calculation. None of these factors is inherently incriminating. But when aggregated through Palantir’s algorithm, they create a profile that ICE uses to prioritize who gets deported and when. The Brennan Center for Justice on ImmigrationOS has more technical detail on how this scoring works, but the practical effect is clear: the system automates the identification of enforcement targets, removing human judgment from the initial stage of deportation decisions.
What troubles me most is the opacity. When I called ICE’s public affairs office to ask which specific data points contribute to an individual’s risk score, I was told that information is classified. Local police chiefs I spoke with said they don’t have access to the algorithm’s logic either. They’re being asked to share data without understanding how that data will be used to identify people in their communities.
The Cities Fighting Back, and What’s at Stake
In February, Chicago, Los Angeles, and Denver filed a joint amicus brief in the 7th Circuit Court of Appeals directly challenging the federal government’s authority to mandate data-sharing with local law enforcement. I called the attorney handling Chicago’s part of the case, and she walked me through their legal theory: the federal government, they argue, cannot use conditional funding to conscript local police into federal immigration enforcement in ways that violate Fourth Amendment protections.
The brief focuses on a specific problem. When ICE uses ImmigrationOS to generate a list of enforcement targets and then shares that list with local police departments, it’s effectively using local law enforcement to do federal work. Unlike federal agents, local cops operate under state privacy laws that often provide more protection than federal law does. By routing immigration enforcement through local police, the argument goes, ICE circumvents state-level privacy protections that would otherwise limit what kind of data can be collected and how it can be used.
I spent several hours reviewing discovery documents filed in the case, and the scope of data sharing is staggering. One police department reported sharing over 4.2 million records with ICE in a single year. That includes traffic stops, utility account information, and vehicle registrations. The city’s argument is that this level of data integration essentially converts local police into immigration agents, which changes their relationship with the communities they serve.
The Mechanics of Information Flow and What Gets Lost
What interests me most, methodologically, is how information transforms as it moves through this system. A driver’s license application that started as a state document gets pulled into a federal database. A utility payment record originally collected for billing purposes becomes an enforcement data point. A social media post that someone made publicly gets fed into a risk algorithm they don’t know exists. Each transformation changes the original meaning of the information.
I’ve spent years reading municipal budget documents and city council agendas, and I can spot when governments are trying to hide something. What I’m seeing now is more subtle. Police departments aren’t trying to hide their participation in ImmigrationOS. They’re being told to participate by federal funding mandates, and they’re not entirely sure what’s being done with the information they’re sharing. The information flows in one direction, toward ICE, but the feedback loop runs the other way: federal money keeps flowing to local police only if they continue sharing.
This isn’t a simple surveillance story. It’s a story about how federal law enforcement shapes local governance through financial pressure, how opaque algorithms become enforcement tools, and how information collected for one purpose gets repurposed for another without anyone’s knowledge or consent.
If you’ve had direct experience with ICE data-sharing practices in your city, or if you’re a local official trying to navigate these federal funding conditions, I want to hear from you. The lines are always open, and I verify everything three times before it becomes a story.