Federal AI and Algorithmic Monitoring Programs (2020s–ongoing)

• Authority: No single law. Operating under existing agency authorities — often without specific statutory authorization.
• Congressional vote: None for the specific programs. General appropriations fund agencies that deploy them.
• Status: Active and expanding. DHS/Palantir AI platform: $1 billion contract signed February 2026.

There is no AI Surveillance Act. Nobody voted to let an algorithm decide your risk score. Nobody debated whether a machine should recommend denying your disability claim or flag you for extra screening at the airport based on your data profile. The federal government is deploying AI systems that make decisions about millions of Americans' lives — health, finances, travel, immigration, benefits — and nobody asked Congress, and Congress has not insisted on being asked.

The $1 billion Palantir contract signed in February 2026 is the clearest marker. DHS is building the largest domestic AI surveillance platform in US history, combining purchased location data, social media monitoring, biometric databases, and immigration records. No statutory authority. No public debate. No accountability mechanism. Just a procurement contract and a billion dollars of your money.

How It Passed

It did not pass. There is no law. That is the crisis.

Federal agencies are deploying AI surveillance and decision-making systems under general statutory authority that was written long before AI existed. The Privacy Act of 1974 requires agencies to publish notices about data systems — it was not written to govern AI. The Administrative Procedure Act requires agencies to follow notice-and-comment rulemaking — but agencies are classifying AI deployments as procurement decisions, not rulemaking, to avoid public scrutiny. The Bank Secrecy Act authorizes financial monitoring — it did not contemplate AI systems that integrate location, biometric, and social media data with financial records.

Congress has considered AI regulation. Multiple bills have been introduced — the American Data Privacy and Protection Act, various AI accountability proposals, the Algorithmic Accountability Act. None have passed. The tech industry lobby and the surveillance industry lobby are well-funded and patient. The bills stall in committee. The AI systems get deployed anyway.

The DHS/Palantir contract is the most recent and most stark example. DHS awarded Palantir Technologies a $480 million contract extension in 2023 and a new $1 billion contract in February 2026 to build an AI-integrated analytics platform across all DHS components — including ICE, CBP, TSA, USCIS, FEMA, and the Secret Service. The contract covers integration of commercially purchased data broker records — location history, consumer profiles, social media data — with government biometric and immigration databases, all processed through Palantir's AI analytics engine.

No specific statutory authority. No congressional authorization for this specific use. No meaningful public debate. A procurement contract awarded to a company co-founded by Peter Thiel, whose relationship with government surveillance is extensive, using authority derived from appropriations bills that fund DHS operations generally.

What It Does To You

The DHS Automated Targeting System (ATS) assigns risk scores to travelers — every person entering or leaving the United States, and potentially domestic travelers as well. The system uses commercial data broker records, travel patterns, financial data, and government databases to compute a score that determines how much scrutiny you receive at a checkpoint. You are not told your score. You are not told what data was used. You cannot correct errors. You can be flagged for additional screening — or placed on a list that affects your ability to travel — based on the output of an algorithm you have never seen, using data you may not know the government holds about you.

The Social Security Administration uses algorithmic tools to screen disability claimants. The algorithm flags cases for additional review or recommends denial based on scoring models. Disability applicants — often people with serious health conditions, facing financial crisis — have their claims evaluated by systems they cannot see, using criteria they are not told, with no right to know an algorithm was involved in the decision.

The IRS uses algorithmic audit selection systems that score tax returns. The algorithm determines who gets audited. Research from the Treasury Inspector General has documented that low-income taxpayers claiming the Earned Income Tax Credit are audited at higher rates than wealthy taxpayers — an outcome that is at least partly driven by algorithmic selection tools that treat easily-verified income sources as easier targets than complex financial instruments.

DHS/CBP border processing uses automated analysis of social media data and travel records to assess risk scores for travelers at ports of entry. Officers have access to these scores when making detention and inspection decisions. The methodology is not public. The data sources are not fully disclosed.

ICE uses AI analytics — specifically tools built by Palantir — to identify, locate, and prioritize enforcement targets. These systems integrate commercial data broker purchases, DMV records, utility records, and social media monitoring. Individuals are flagged for enforcement action based on algorithmic scores without individualized judicial authorization.

Rights It Strips

Your due process right to know what decision is being made about you and why. When an algorithm denies your disability claim, flags your tax return for audit, assigns you a border risk score, or marks you for immigration enforcement, you typically do not know the algorithm was involved. You do not know what data it used. You cannot challenge the algorithmic decision because you are not told an algorithmic decision was made. The government is making consequential decisions about your life using processes designed to be invisible.

Your right to correct government records about you. The Privacy Act gives you the right to access and correct government records about you. AI systems ingest thousands of data points from commercial and government sources. The input data is often wrong — data brokers have documented error rates of 30-50% for some data categories. When an AI system uses incorrect data to compute a risk score that affects your life, there is no meaningful correction mechanism. You cannot identify the error you cannot see.

Equal protection under the law. NIST's comprehensive facial recognition study (published 2019) documented that leading commercial facial recognition algorithms had error rates up to 100 times higher for Asian and African-American faces than for white male faces. Predictive policing algorithms trained on historical arrest data encode historical patterns of over-policing in minority communities and amplify them. When the government deploys biased AI systems to make law enforcement and benefits decisions, it is administering a discriminatory system at scale — with the discrimination laundered through algorithmic opacity.

Your right to human review of consequential government decisions. No federal law requires a human reviewer to make the final decision when AI recommends denial of your benefits, flags you at a border crossing, or scores your tax return for audit. The human may be present, but they are processing algorithmic outputs from systems they do not fully understand, with insufficient time for genuine independent review.

Documented Abuses

The bias documentation is extensive and official.

NIST Facial Recognition Technology Evaluation (2019): Federal testing of 189 facial recognition algorithms found that 99 of the algorithms showed higher false positive rates for African-American women compared to white women — some by factors of 10 to 100. These are not allegations from advocacy organizations. These are findings from the federal agency responsible for testing federal AI tools. The government tested its own AI, confirmed profound racial bias, and continued deploying it.

IRS audit disparity: A 2023 Stanford study found that Black taxpayers were audited at 2.9 to 4.7 times the rate of non-Black taxpayers with similar incomes. The IRS acknowledged the disparity and attributed it partly to algorithmic audit selection focused on EITC returns — a category disproportionately claimed by low-income Black taxpayers. The IRS committed to studying reforms. The algorithm continued operating.

Predictive policing in Chicago and Los Angeles: Documents obtained through public records requests showed that Chicago's Strategic Subject List — an algorithmic score that ranked individuals by "risk" of future violence — had documented errors, included people who had died, and was used to justify preemptive police contact. Los Angeles suspended its predictive policing program in 2020 after LAPD records showed that the algorithm's predictions were no better than chance in many categories, and that Black and Latino neighborhoods were disproportionately targeted.

TSA screening list errors: Government Inspector General reports have documented cases of individuals placed on TSA watchlists in error, including children and people with names similar to those of actual suspects, who experienced years of stigmatizing additional screening with inadequate redress processes.

The Palantir ICE contracts: ACLU and investigative journalism documentation of Palantir's ICE deployments showed the system was used to identify and arrest individuals with no criminal records beyond civil immigration violations, using data aggregated without their knowledge from utility companies, school records, and commercial databases. These enforcement targets had no notice that their data was being compiled, no opportunity to correct errors, and no meaningful appeal before enforcement actions were initiated.

Who Pushed This

Palantir Technologies and the defense/surveillance contractor ecosystem. Palantir's entire business model is government contracts for AI-integrated surveillance and analytics. The company's revenue depends on expanding the scope of government AI deployment. Peter Thiel, co-founder and early investor, has deep relationships with both Republican and Democratic government officials. Palantir has donated to campaigns on both sides of the aisle and embedded former government officials throughout its ranks. The revolving door between government surveillance agencies and Palantir is well-documented.

Agency officials who prefer operational opacity. ICE, CBP, TSA, and other enforcement agencies prefer AI systems that make their decision-making harder to challenge. If an officer denies you entry, you can challenge their judgment. If an algorithm scores you as high-risk, the challenge is far more difficult — the methodology is proprietary, the data is undisclosed, and the legal framework for contesting algorithmic government decisions is underdeveloped.

Appropriations committees that fund without oversight. Every year, Congress appropriates money for DHS, SSA, IRS, and other agencies that deploy these systems. The appropriations bills do not include specific conditions on AI deployment. The oversight committees do not require transparency in AI procurement. The money flows, the contracts are awarded, and the systems expand — while Congress debates other things.

The data broker industry. Commercial data brokers — companies like LexisNexis, Acxiom, and dozens of others — sell the underlying data that feeds federal AI surveillance systems. Their business model depends on the government buying their products. They lobby against data privacy legislation that would restrict commercial data collection. The intersection of commercial data commerce and federal AI surveillance is the infrastructure layer of this entire problem.

Key Votes

There are no votes. And unlike the GTO situation, where a single administrative program has expanded quietly, the absence of votes here covers an entire category of government activity — AI-based decision-making and surveillance — that is expanding across every major federal agency.

The accountability record is built from:

• Who has refused to hold hearings on the Palantir/DHS contract despite its $1 billion scope and civil liberties implications
• Who sits on the oversight committees — House Judiciary, House Homeland Security, Senate Intelligence, Senate Judiciary — and has not demanded public reporting on algorithmic decision systems
• Who voted to fund agencies without requiring AI accountability provisions in appropriations
• Who has taken campaign contributions from Palantir, defense contractors, and data broker companies and remained silent as these systems expanded
• The 218+ House members and 51+ senators needed to pass an Algorithmic Accountability Act — and why that majority has not been assembled after years of proposals

Why This Matters for We The Citizens

Federal AI surveillance programs represent the cutting edge of the accountability crisis We The Citizens exists to address. These are not programs that can be easily reversed with a single vote. They are infrastructure — contracts, databases, platforms, integrated data flows — that will be far harder to dismantle than they were to build.

Every other bad law on this site can be traced to a congressional vote. A vote that passed. A vote that failed. Members whose names are on the record. PADFAA: 353–36. PAFACA: 360–58. The PATRIOT Act: 357–66 in the House, 98–1 in the Senate. Even the Bank Secrecy Act has a history, a vote, a signed bill.

The AI surveillance programs have none of that. They were built in procurement offices, funded through general appropriations, authorized under statutes written for different purposes, and expanded by agency determination. No member of Congress has ever had to vote yes or no on whether the federal government should build a $1 billion AI surveillance platform integrating commercial location data with immigration and biometric records.

That is the accountability gap. We The Citizens' task is to make the invisible visible: who is funding these agencies without accountability conditions, who is sitting on oversight committees and looking away, who is taking money from the surveillance industry, and who — if anyone — is actually fighting to require transparency and accountability in government AI.

The AI accountability movement needs to succeed in the next Congress or the infrastructure becomes permanent. Once a $1 billion surveillance platform is built, integrated, and operationalized across every DHS component, it will be defended as essential infrastructure. The window to establish meaningful oversight is now. We The Citizens should be tracking every member who could be a vote for accountability and isn't.

See also: Bad Laws Overview