The IRS Paid a Tech Company $130 Million to Hunt Tax Cheaters. Here's What That Means

The IRS has spent more than $130 million paying Palantir Technologies to analyze government databases in search of tax fraud and financial crime, according to reporting from The Intercept. The company has been examining dozens of different pools of government data about Americans to support tax investigations — one of the biggest partnerships between a military-focused tech firm and a civilian tax agency.

Palantir started working with the IRS in 2014, according to government contracting records reviewed by Wired. What began as a simple tool has grown into a much larger system that touches both criminal investigations and routine tax enforcement.

A New Way to Pick Targets

The heart of this partnership is a tool called SNAP — Selection and Analytic Platform. The IRS built it with Palantir to do something that sounds simple but is actually quite hard: figure out which tax returns deserve closer inspection.

For decades, the IRS used a formula called the Discriminant Information Function, or DIF. It was a statistical method — basically a checklist of mathematical rules — that flagged tax returns that looked suspicious. That system still works, but it has limits.

Before Palantir came in, the IRS was running more than 100 separate computer systems and 700 different methods, built up over many years, just to select which cases to investigate. SNAP brought all that information together into one place. Now the system uses advanced software to spot patterns — like networks of fake companies, or suspicious money transfers — that older methods might miss.

In 2023 alone, the IRS spent $1.8 million improving SNAP's ability to pick cases for audits. IRS documents show that the new Palantir system has replaced older tools the agency had relied on.

Building a New Department

The IRS has now created a formal unit dedicated to these advanced analytics capabilities. It sits within the Criminal Investigation division and is part of what officials call an "Advanced Analytics and Innovation strategy."

This unit works on two kinds of cases. Some are investigations that start with a tip or a complaint. Others are proactive — the system runs through financial data looking for patterns that suggest someone might be breaking tax laws. It looks at transaction histories, cross-checks them against other government records, and uses a risk-scoring system to identify potential targets.

The scale here matters. We have seen defense contractors move from military work into civilian government before — most notably after 2001, when federal surveillance expanded significantly. Palantir developed its data-analyzing techniques for military and intelligence operations overseas. Now those same techniques are being applied to domestic financial records about ordinary Americans.

Who Cares, and Why

This partnership has raised concerns among transparency advocates and government watchdog organizations. Palantir's founder, Peter Thiel, was a major donor to President Trump's 2016 campaign and worked closely with his administration. The New York Times has previously reported that Palantir helped the Trump administration try to share data more easily across different federal agencies.

Chioma Chukwu, director of the government watchdog group American Oversight, expressed concerns about consolidating so much financial information into one system, according to The Intercept. Her organization has filed legal requests against multiple federal agencies, including the IRS, trying to find out how Palantir's systems are actually being used to track and analyze people's financial lives.

The timing of this expanded partnership is worth noting. The IRS received billions of dollars in new funding through the Inflation Reduction Act, with some money specifically earmarked for better enforcement and modernizing technology. This partnership expanded during that same period.

What This Means Going Forward

The shift to algorithmic case selection is real. Instead of relying on rules written by humans, the IRS is now using machine learning — software that learns patterns from historical data — to identify cases worth investigating. The computer systems can spot connections across huge amounts of data that would be nearly impossible for a person to find by hand.

The practical upside is significant. The IRS's enforcement capabilities, particularly for complex financial crime, could improve. These systems can spot shell company networks and coordinated tax avoidance schemes that traditional methods might miss. That matters, because organized tax fraud often requires examining connections across multiple companies and jurisdictions.

However, there is a legitimate tension worth addressing. Consolidating dozens of federal databases under one analytical platform creates an unprecedented view into how Americans manage their money. The public debate about what limits should apply to this kind of surveillance capability, and how to ensure it is not misused, has been limited. These are decisions with real consequences for privacy and trust, and they deserve more public visibility.

The relationship between the IRS and Palantir is likely to grow further as the agency continues modernizing. It may eventually include additional data sources and capabilities beyond current tax enforcement work. This reflects a broader trend in which specialized defense contractors increasingly build the data-processing systems that civilian agencies rely on. Understanding how these partnerships work, and what safeguards exist, is important for anyone concerned about both effective government and personal privacy.