When longtime Indiana politician Sen. Jim Banks (R-IN) lobbed a powerful allegation against the Census Bureau last October, what had been a low-profile, fringe GOP conspiracy theory burst to the fore. Banks said he believed the Census Bureau had injected errors in the decennial population count and survey which were so severe, they netted Democrats more representation in Congress at the expense of Republican seats.
“The Biden admin used a shady ‘privacy’ formula,” Banks wrote on X, adding that the 2020 census “scrambled the data and miscounted 14 states. It included illegal immigrants and handed Democrats extra seats.”
In a letter to Commerce Secretary Howard Lutnik that Banks posted on X, the senator repeatedly asserted that the use of a privacy technique called noise infusion, rolled out to help the Census Bureau better protect the individual identities of respondents, resulted in population count errors that interfered with the process of allocating political representation in Congress, known as apportionment. Those errors overwhelmingly supported blue states and hurt red ones, said Banks, who also conflated the use of the technique with myths about noncitizen voting.
His post has more than 9,000 reposts and has, according to X, been viewed more than 8.5 million times. Becoming a rallying cry for Trump allies and MAGA-aligned influencers against Democratic “fraud,” Banks’ post was quoted by far-right provocateur Mike Cernovich, known for promoting the fringe conspiracy theory known as Pizzagate, to his 1.4 million followers.
Peter Bernegger, a staunch election denier with more than 100,000 followers, also pushed it, writing in a lengthy post that the “2020 Census was manipulated via a secret algorithm.”
The editorial board at the right-leaning Washington Times wrote in October that the Census’ use of noise infusion, a type of differential privacy, “formalized the publication of fraudulent information.”
The problem is, Banks’ viral post wasn’t true. It was part of a disinformation campaign using categorically false information about census administration that has reduced a highly advanced statistical technique to just another arm of the disproven, right-wing delusion that immigrants are somehow enabling Democrats to rig federal elections and claim undeserved political power. More extreme versions of the claim allege that privacy protections used by the Census Bureau distort apportionment — the process of using decennial census data to determine how many congressional seats each state receives. It overlaps with hardline GOP immigration policy, as Trump officials and allied groups imagine huge numbers of immigrants have been wrongly imputed to blue states. Less extreme versions of the claim focus instead on individual congressional districts and redistricting. Ending the use of differential privacy, both wings assert, will alleviate this unfair blue state advantage.
Experts say both versions are baseless.
“The business of getting rid of differential privacy will have no impact on anything political,” Steven Ruggles, a professor at the University of Minnesota and a creator and director of IPUMS, the world’s largest population database, told TPM. “Nothing regarding differential privacy or swapping could possibly affect apportionment. It’s just much too subtle; very small adjustments to the data.”
Banks’ office did not respond to a TPM request for comment.
The Census Bureau is legally required to protect the identities of people who respond to its surveys. In 2020, Bureau employees accomplished this by using noise infusion, which added and subtracted information from individual respondents’ data at granular geographic levels to fuzzy the figures and keep bad actors from re-identifying people who filled out the census.
Differential privacy isn’t applied at all to the state population totals that are used to determine congressional representation by state, experts told TPM. And the amount of “noise” eventually added to fuzzy individual demographic data is too small to impact redistricting, as Republicans claim.
“The odds that [differential privacy] could ever affect a majority-minority voting district is also very implausible,” said Ruggles, who had once been an outspoken critic of the 2020 Census roll-out of differential privacy. “It’s sort of about the likelihood of an election being determined by one vote.”
Nevertheless, the idea that “covert” census policies help Democrats animates nearly everything about how Trump’s reactionary movement sees its crusade against the Census Bureau, which it has targeted as part of its larger mission to change how the population is counted — and help cement Republican legislative power. It is likely to play a crucial role as it works to fulfill the president’s demand for a whole “new” census in time for 2030.
TPM set out to figure out where these ideas about differential privacy came from. Some of the research that underlies this reporting was conducted by Jae June Lee, a privacy researcher at Cornell University, and shared with TPM.
Our interest in this conversation was piqued because Trump’s Census Bureau appears to be acting on these unfounded beliefs. In recent months, the administration took a series of unprecedented actions aimed at changing how the agency handles privacy, apparently in hope of shrinking Democrats’ political power.
In June, it banned a set of policies for protecting Americans’ privacy, including differential privacy — policies that had been carefully fine-tuned over the last several years — seemingly without any expert input.
Then, TPM uncovered a draft Census Bureau rule circulating within the administration that sought, among other things, to exclude many immigrants from the population count used for apportionment, an attempt to redistribute that violates the 14th Amendment requirement for the census to count all whole persons in the country. It also sought to end the collection of most demographic information, citing concerns about privacy.
Underlying each of these developments is the right-wing, conspiratorial belief that the Census Bureau’s handling of privacy somehow favors Democrats in redistricting and apportionment. That idea, TPM found, was having a moment back in 2021.
In March of that year, the state of Alabama filed a lawsuit, which it would later lose, challenging the Census Bureau’s use of differential privacy and calling for a complete re-do of the 2020 Census. Two months later, the Claremont Institute, a conservative think tank, published an article critiquing the technique written by a future Project 2025 contributor who worked at the Census Bureau during Trump I.
Before differential privacy was used in the 2020 Census, experts like Ruggles and others were concerned about its impact on data accuracy, which resulted in public debates and data demonstrations. After years of tests, Ruggles said he was satisfied that the amount of noise was so miniscule, it wouldn’t compromise data quality. Today, right-wingers seem to be seizing on that controversy, contorting early, valid concerns about differential privacy into an ongoing, devious and misleading political campaign.
With these early challenges to differential privacy as a foundation, the issue reemerged after Trump was reelected.
On Aug. 7, 2025, Trump, in a post on the social media platform he owns, Truth Social, directed the “Department of Commerce to immediately begin work on a new and highly accurate CENSUS.” That census, Trump wrote, should exclude undocumented immigrants.
The same day, the Center for Renewing America — a think tank founded by now-Office of Management and Budget Director Russ Vought — published a 6,000 word report called “The 2020 Census Fraud: How Differential Privacy Skewed Representation.” The Census Bureau’s differential privacy use “tilted the political scales favorably toward Democrats for redistricting purposes,” the report said, in opposition to widely accepted professional understanding about differ