Reading · Judith Dayal · RSS Amplifier

The Trump regime has deleted or altered at least 366 federal data collections since returning to power, including information about police misconduct, drug overdoses, violence against children, air pollution, worker pay, education, immigration, veterans and the health of LGBTQ Americans.

Twenty-eight datasets have been terminated entirely. Another 338 have been modified. The changes stretch across more than 60 federal agencies, according to a new tracker created by dataindex.us, a group of researchers, developers and data-policy experts who have spent the past year trying to document what is disappearing from the federal government.

That number is worth sitting with for a minute, because government data can sound like one of those painfully boring things that lives in a spreadsheet somewhere and matters mostly to people who enjoy spreadsheets. In reality, we build enormous parts of our lives around it without ever thinking about where it comes from.

Federal health data tells public-health departments where people are overdosing. It tells researchers which populations are being diagnosed with HIV, where diseases are spreading, which communities are experiencing higher rates of illness, and whether certain groups of Americans are dying at rates they shouldn't be. It helps us track maternal and infant health, mental health, infectious disease, cancer, drug use and violence. This is information doctors, researchers and public-health officials use to figure out what is happening to us and where help is needed.

And health is only the largest category. The data being altered or eliminated also covers education, veterans, immigration, labor, environmental hazards, police misconduct and violence against children. These aren't numbers collected for the hell of it. They are part of how we know whether people are safe, whether policies are working and where something has gone terribly wrong.

One of the datasets that disappeared was the National Law Enforcement Accountability Database, a federal database containing thousands of records of serious misconduct by federal law-enforcement officers. It had been created so agencies could identify misconduct when making hiring, promotion and assignment decisions. By the time it was taken offline in February 2025, participating agencies had reported thousands of misconduct records.

The Justice Department shut it down after Trump rescinded the executive order that established it.

Another affected program is the Drug Abuse Warning Network, which collected information from hospital emergency departments about drug-related visits. That kind of information helps researchers and public-health officials see what drugs are hurting people, where problems are emerging and how patterns are changing. The tracker now lists the program as terminated.

There is also a federal survey designed to measure physical, sexual and emotional violence against children and young people. There were pollution readings collected from air-quality monitors at U.S. embassies around the world. There was information about farmworker wages. Teacher pay. Workplace conditions. Environmental hazards.

Those modifications are especially important because a dataset does not have to disappear entirely to become less useful. Change what information the government collects, stop asking a particular question, or remove a group of people from the categories being measured, and eventually the historical record changes with it.

That is exactly what has been happening to federal data about LGBTQ Americans.

On Trump's first day back in office, he signed an executive order declaring that the federal government would recognize only two sexes and directing agencies to remove forms, policies, communications and other materials that the regime considered to promote what it calls “gender ideology.” The order specifically instructed federal forms not to request gender identity.

The Williams Institute at UCLA School of Law spent a year tracking those changes and found that approximately 360 federal data collections removed at least one measure of sexual orientation or gender identity between January 2025 and January 2026. Gender-identity measures were removed from 338 collections. Sexual-orientation measures were removed from at least 60.

In other words, most of these changes did not go through the usual public notice-and-comment process that would have given people an opportunity to see what the government was changing and respond to it. Researchers at the Williams Institute found that only a small fraction went through the formal process.

The CDC's Behavioral Risk Factor Surveillance System has been used to study the health of Americans across the country. Gender-identity information has been removed from federal health collections. Other changes affect data used to study HIV, intimate-partner violence and sexual violence.

Imagine that five years from now researchers want to know whether transgender Americans are experiencing a particular health disparity. Or whether a certain population is being disproportionately assaulted. Or whether suicide rates have changed. Or whether a public-health program actually worked.

When a population disappears from the data, researchers lose the ability to reliably measure what is happening to that population. Policymakers lose evidence they could use to allocate resources. Journalists lose evidence they could use to hold the government accountable. And the public loses one of the ways we can prove that a problem exists.

The effects do not necessarily show up the day a question is deleted. Existing research is still there. Old datasets still exist in archives. Researchers and librarians have been racing to preserve federal information before it disappears.

Every year that information is no longer collected creates another hole. Eventually researchers trying to compare 2024 with 2027 or 2030 encounter a point where the government simply stopped asking the same questions. Trends become harder to establish. Changes become harder to measure. People become harder to see.

And this extends well beyond LGBTQ data.

Earlier this summer, a Guardian investigation found that more than 1,600 Department of Energy webpages had disappeared, including pages containing information about energy efficiency, utility costs and protecting the electrical grid during extreme heat. Other federal information removed or disrupted since Trump returned to office has involved climate risk, maternal health, hunger and hazardous chemicals.

There are now librarians, researchers and volunteer archivists around the country trying to save government information before it disappears. The Data Rescue Project has been preserving federal datasets. Other volunteers have begun photographing signs and historical material in national parks because they are worried that information there will be changed or removed.

Americans are trying to make backup copies of their own government because they no longer trust the government to preserve an accurate record of itself.

Federal data belongs to the public. Taxpayers paid to collect it. Scientists, doctors, universities, local governments, businesses, journalists and ordinary people have relied on it for decades. Much of it was gathered precisely because no individual researcher or organization could possibly collect information on the scale of the United States government.

And once a government has collected years or decades of consistent information, breaking that continuity can cause damage that cannot simply be repaired by putting a webpage back online later.

You can restore a deleted file.

That is what makes this so chilling to me. Governments make decisions based on information, but information also gives the rest of us a way to judge the decisions governments make. We can see whether a policy reduced deaths. We can see whether pollution increased. We can see who is being arrested, who is getting sick, who is earning less, who is experiencing violence and who is being left behind.

And a government willing to erase parts of that paper trail gains something extremely valuable: fewer measurements against which its version of reality can be tested.

Denice Ross, a former U.S. chief data scientist and one of the people behind dataindex.us, warned that the country is reaching a point where we may simply stop having some of the information necessary to make decisions that save lives.

I think there is an even more basic way to understand what is happening.

And if enough of the public record disappears, eventually the government gets to tell us what is happening without leaving us nearly as much evidence to check for ourselves.

As always, please like, comment, and share this post if you can. It really does help push my work through the algorithm and get stories like this in front of more people. And this one, especially, feels like something people need to see.

The Government Is Erasing the Data It Doesn’t Want You to See

Comments