The Department of Homeland Security uses AI tools for immigration enforcement and surveillance—including license plate readers, deportation targeting apps, and facial recognition—but avoids transparency requirements by claiming humans independently review AI outputs. However, the article argues this defense ignores how AI inherently influences human decision-making, allowing DHS to exploit a loophole in the Trump-Vance administration's AI governance memo that requires safeguards only when AI is the 'principal basis' for decisions affecting rights and safety.