For years, little was known about the multibillion-dollar company that handles data for the US immigration enforcement agency. Now, a cache of emails, training documents and reports sheds light on how Palantir helps Ice with investigations and on-the-ground enforcement

For years, little was known about how Ice uses Palantir’s technology. The company has consistently described itself as a “data processor” and says it does not play an active role in any of its customers’ data collection efforts or what clients do with that information.

Now, a cache of internal Ice documents – including hundreds of pages of emails between Ice and Palantir, as well as training manuals, and reports on the use of Palantir products – offer some of the first real-world examples of how Ice has used Palantir in its investigations and during on-the-ground enforcement operations.

The documents, which were obtained by immigrant legal rights group Just Futures Law through a Freedom of Information Act request and reviewed by the Guardian, largely cover Palantir’s contract with Homeland Security Investigations (HSI), the investigative arm of Ice that is responsible for stopping the “illegal movement of people, goods, money, contraband, weapons and sensitive technology”.

Palantir recently won a $30m contract to build the government a new platform called ImmigrationOS that will service Ice branches beyond HSI, and aims to “streamline” the identification and deportation of immigrants. While the documents are largely limited to HSI’s operations, it’s easy to see how the tools could be of use to other branches of Ice.

One document showed that agents could use Falcon to search for people’s names, known locations, vehicles or passport information against Department of Homeland Security (DHS) and other federal databases like the Enforcement Integrated Database (EID) – a vast database that holds biometric and personal information on anyone who has been encountered or arrested, detained or deported by any DHS agency – on the go.

In addition to collecting and analyzing field interviews, the agents used Falcon for “blueforce tracking”, a military term for tracking someone’s GPS location. Training documents show Ice agents could also use Falcon to track a person’s location, including their “route and movement”, hour by hour using cell phone tower records.

  • Basic Glitch@sh.itjust.worksOP
    3·
    1 day ago

    Experts warn that, in light of the Trump administration’s threats to crack down on “far-left” groups, the scope of application for Palantir’s tools could only grow. Already, a new report from the Intercept revealed HSI subpoenaed and received information from Google about international students who were being investigated over their pro-Palestinian activism.

    “Now [with access to more federal databases] Ice can use this type of surveillance apparatus on anyone – not only anyone who is undocumented but anyone who this administration wants to criminalize and anyone who the administration wants to put under surveillance,” said González of MediaJustice.

    This combined with live facial recognition camera networks and drones already being proposed in some cities is what makes it such a scary tool with so much potential for abuse.

    Imagine the way people have been unfairly targeted and harassed in the past for political affiliations. Like think about the level of spying that was done to MLK and his family, and imagine that the FBI had access to something like this, and could use it against anyone in the entire country.

    There is also increasing overlap between federal and local agencies sharing these platforms. Soon even local police may be able to access these networks and databases, and use them to harass anyone for any reason, even personal. Like imagine you or your family member end up on the wrong local cop’s shit list, and they decide to target you.

    The most concerning is the thought of all of this combined with the tech like Lavender being used in Gaza. Inescapable 24/7 AI scanning, monitoring, and surveillance that can track you anywhere, and send a drone strike to take you out. (As well as anyone unfortunate enough to be in your vicinity at the time.) And sometimes the AI just fucks up and ends up taking out an incorrect target.

    Israel still hasn’t revealed what their "secret weapon was, but I assumed those nuclear scientists in Iran that were all killed simultaneously in their sleep (Operation Narnia) were targeted using something similar to the tech Palantir already provides for the attacks on Gaza.

    The Integration of Artificial Intelligence in Targeted Operations: An Analysis of Israeli and US Involvement in Iran

    US Policy on Lethal Autonomous Weapon Systems (LAWS) DoD Directive 3000.09 defines LAWS as weapon systems that can select and engage targets without further human intervention once activated (“human out of the loop” or “full autonomy”). The directive mandates that all systems, including LAWS, must be designed to enable "appropriate levels of human judgment over the use of force.

    Accountability and Human Control in Autonomous Systems A critical challenge posed by AWSs is the “accountability gap.” The increasing autonomy and opacity of AI systems in lethal operations create a profound difficulty in assigning responsibility for errors or violations of international law. Since machines lack intent or moral reasoning, assigning criminal responsibility for errors or violations of international law becomes exceptionally difficult. This challenge is compounded by the complex chain of command involving developers, deployers, and operators , often shielding those lowest in the chain.