Guardian alleges 400 dataset alterations

The Guardian reported on August 18 that the Trump administration had deleted or altered nearly 400 U.S. federal datasets since January 2025, citing a new tracker maintained by dataindex.us. The count behind the headline was 28 datasets deleted and 338 data collections modified across more than 60 agencies, according to the tracker and accounts that republished the Guardian report. (centerforpoliticalawareness.com) The story resurfaced on social media on September 8, but the underlying report is from mid-August, not from Tuesday. (dataindex.us) ### Where does the “400 datasets” number come from? Dataindex.us says its Federal Data Terminations Tracker is a “verified, regularly updated record” of federal datasets, data elements and tools terminated or removed since January 2025. (centerforpoliticalawareness.com) The version cited in August counted 28 deleted datasets and 338 modified collections, a total often rounded in coverage and social posts to “nearly 400.” (captaincompliance.com) The tracker says it grew out of a crowdsourced archive called “Dearly Departed Datasets” and combines community reporting with expert review. That matters because the allegation is not based on a single agency announcement; it is based on a running catalog assembled by outside researchers monitoring changes across the federal data system. (centerforpoliticalawareness.com) ### What kinds of federal data were reported missing or changed? (dataindex.us) The Guardian article, as reproduced by other outlets, said the affected records spanned environment, labor, law enforcement, education and health. One example cited was a database of federal law enforcement officer misconduct records. (captaincompliance.com) Another was a survey measuring physical, sexual and emotional violence against children and youth. (dataindex.us) Health data appeared most affected, accounting for about 40% of the terminated or altered collections tracked by dataindex.us, according to summaries of the Guardian report and follow-up coverage. (dataindex.us) Becker’s Hospital Review, citing the same tracker, also said health records made up the largest share of the changes. (centerforpoliticalawareness.com) ### Does “altered” mean the same thing as “deleted”? Dataindex.us distinguishes between datasets that were terminated and data elements that were removed or changed. That means the headline number combines outright disappearances with narrower revisions to fields, categories or collection methods. (centerforpoliticalawareness.com) One reported example involved the removal of information on queer and trans people from some health datasets, which coverage linked to Trump’s executive order on “gender ideology.” In other words, the allegation is broader than webpages going dark; it also includes changes inside surviving datasets. (beckershospitalreview.com) (rsn.org) ### Why did this circulate again on September 8? An X post on September 8 cited the Guardian piece and framed it as a current political talking point. (dataindex.us) The post appeared alongside unrelated commentary about Florida politics, including Byron Donalds, but those are separate claims and not part of the Guardian dataset report. (edzardernst.com) The clearest way to read the timing is that social media revived an August 18 report on September 8. (rsn.org) The underlying sourcing still points back to the Guardian story and the dataindex.us tracker, not to a new government filing released Tuesday. (dataindex.us) (edzardernst.com) ### What can readers check next? The Federal Data Terminations Tracker remains live at dataindex.us and says it is updated on an ongoing basis. Readers looking for the underlying entries can start there, where the project says it records terminations and removals since January 2025 and accepts submissions at removals@dataindex.us. (edzardernst.com) (dataindex.us) (centerforpoliticalawareness.com)