Robert F. Kennedy Jr. on the Joe Rogan Experience #2461, fact-checked

What the evidence shows: CMS reported combined FY2023 improper payments across Medicare (Fee-for-Service, Part C, and Part D) and Medicaid of roughly $101.5 billion ($31.2B plus $16.6B plus $3.4B plus $50.3B), a figure in the range Kennedy cites. But CMS explicitly distinguishes improper payments from fraud: most stem from insufficient documentation or administrative errors rather than proven intentional deception, and in FY2023, 82% of Medicaid improper payments were attributed to insufficient documentation rather than confirmed fraud or abuse. Federal agencies do not publish a reliable annual estimate of fraud specifically, since fraud (unlike improper payments) requires proof of intent and is inherently harder to measure. The roughly $100 billion figure is real and well documented, but attributing all of it to "blatant," "industrialized" fraud, as Kennedy does, is not supported by the underlying CMS data, which conflates a broader, mostly non-fraudulent error category with fraud.

What the evidence shows: Durable medical equipment (DME) fraud against Medicare and Medicaid is a well-documented, long-running problem in South Florida, and federal data support the broader pattern Kennedy describes: CMS reported that medical-supply DME suppliers had a 17% revocation rate (roughly triple other supplier types) and imposed a nationwide moratorium on new DME enrollments in February 2026 after linking these suppliers to more than $1.5 billion in suspected fraudulent billing, and the 2026 National Health Care Fraud Takedown charged 455 defendants in cases involving over $6.5 billion in false claims, including DME schemes. However, no primary government report, court filing, or independent news account identifying this specific 129-room hotel, or corroborating that it (or the broader DME fraud operation) was directly run by the Cuban government as opposed to Cuban nationals or Cuban-American individuals acting independently, the pattern documented in decades of prior enforcement actions, could be located. The claim that this specific fraud scheme was directly "run by the Cuban government" goes beyond what any located public record substantiates, making the anecdote unverified as told even though the general category of DME shell-company fraud tied to operators of Cuban origin in South Florida is well established.

What the evidence shows: Los Angeles County has a documented, extreme concentration of hospice fraud indicators: a March 2022 California State Auditor report found the county's number of licensed hospice agencies grew 1,500% since 2010 and that LA County had more than six-and-a-half times the national average number of hospice agencies relative to its aged population in 2019. Subsequent congressional scrutiny in 2026 has cited California, including LA County, for a disproportionate share of hospice-related Medicare billing and estimated taxpayer losses. However, no state or federal audit data supports the specific claim that LA alone has more hospice capacity or providers than the entire rest of the country combined; reported figures put LA County's share as a large but partial fraction of U.S. hospice providers, not a majority. The claim that hospice operations in LA are "all fraudulent" also overstates the audit findings, which identified fraud indicators at a large share of agencies rather than literally all of them. The underlying scale of documented fraud is nonetheless substantial, with officials estimating losses in the hundreds of millions of dollars tied to LA-area hospice schemes.

What the evidence shows: Minnesota's Medicaid-funded autism therapy benefit, known as EIDBI, did grow at the scale Kennedy described: the Department of Justice reported that EIDBI claims "skyrocketed from over $600,000 in 2018 to over $400 million by 2025," a trajectory consistent with Kennedy's account of a program expected to cost a few million dollars a year ballooning to roughly $400 million annually within a few years. As HHS Secretary, Kennedy made these remarks alongside DOJ and CMS officials who in May 2026 announced criminal charges against 15 defendants, including the largest autism Medicaid fraud case ever charged by DOJ, an alleged $46.6 million scheme involving kickbacks to parents and billing for autism services that were not provided or not medically necessary. However, Kennedy's characterization that the program was "all fraudulent, almost" goes beyond what has been proven in court: as of the charges announced, prosecutors had charged specific schemes totaling roughly $90 million in the broader takedown (with the single largest autism case at $46.6 million), not the full $400 million in annual EIDBI spending, and separate Minnesota state reporting has found substantially smaller confirmed-fraud totals than some federal estimates. The dollar figures and program-growth trajectory Kennedy cited are well-supported by DOJ's own account; the claim that nearly the entire $400 million was fraudulent remains an estimate rather than an adjudicated fact.

What the evidence shows: Kennedy claims "overwhelming science" shows acetaminophen use in pregnancy is unsafe and that studies supporting its safety are weak; the largest and most rigorous analysis to date points the other way. A 2024 JAMA study of 2,480,797 Swedish children used sibling-control methods to account for genetic and familial confounding, and found no association between acetaminophen use during pregnancy and children's risk of autism, ADHD, or intellectual disability once siblings discordant for exposure were compared, indicating that associations seen in earlier, less-controlled models were likely attributable to familial confounding rather than the drug itself. This does not support characterizing the evidence as "overwhelming" in the direction of harm, or the safety literature as uniformly weak; the balance of evidence is better described as mixed in uncontrolled observational studies but reassuring in the most rigorous controlled analysis.

What the evidence shows: Kennedy claimed 1970s autism prevalence was under 1 in 10,000 and has since risen to roughly 1 in 31, implying a true epidemic-scale increase. Federal surveillance data show current U.S. autism prevalence among 8-year-olds is about 1 in 31 (3.2%) as of the CDC's most recent Autism and Developmental Disabilities Monitoring (ADDM) Network report, up from about 1 in 150 when standardized multi-site surveillance began in 2000 and 1 in 44 in 2018. No comparable nationwide surveillance system existed in 1970; early low estimates from that era came from small, localized studies using narrow diagnostic definitions and limited case-finding methods, not a national epidemiological survey comparable to the CDC's later network. The CDC and autism researchers attribute most of the measured rise since 2000 to broadened diagnostic criteria (DSM revisions expanding the autism spectrum), increased awareness, more systematic screening, and improved case ascertainment across diverse populations, rather than solely a true increase in underlying incidence. Whether some real increase in incidence has also occurred remains an open, debated question in the research literature, but the magnitude implied by comparing a narrow 1970s clinical estimate directly to modern broad-spectrum surveillance figures overstates the true underlying change. Current status: misleading, the historical comparison conflates methodological changes in case definition and detection with a literal epidemic-scale rise in disease incidence.

What the evidence shows: The 77% figure tracks a Pentagon-derived eligibility statistic: a 2020 Department of Defense estimate that roughly 77% of Americans aged 17 to 24 would be disqualified from military service without a waiver if they applied, up from about 71% in a 2017 DoD estimate. The leading disqualifiers in this line of research are being overweight, drug and alcohol use, and other medical or physical health problems, with a large share of ineligible youth disqualified for multiple overlapping reasons. Kennedy's statement broadens this into a claim about 'American kids' generally, when the underlying estimate applies specifically to a narrower pool of 17-to-24-year-olds who would seek to enlist, and omits that disqualifying categories also include criminal record and, in some versions of the estimate, educational deficiencies. FactCheck.org's review of a near-identical earlier version of this statistic (71%) documents how this Pentagon-derived figure is commonly mischaracterized by public figures -- for instance, described as an applicant-rejection rate rather than an eligibility estimate for the full age cohort, or attributed mainly to education rather than obesity and health. That source confirms the statistic's origin and its history of being oversimplified, though it verifies the pattern and methodology behind this class of figure rather than the specific 77%/2020 estimate Kennedy cites. The core number is broadly consistent with the Pentagon's own published estimates, but Kennedy's framing extends it beyond its actual scope and audience.

What the evidence shows: Meta (Instagram's parent company) disabled Kennedy's personal Instagram account on February 10, 2021, roughly three weeks after Biden's January 20, 2021 inauguration, not 37 hours as Kennedy claimed; a Meta spokesperson said the account was removed for "repeatedly sharing debunked claims about the coronavirus or vaccines." PolitiFact, examining Kennedy's related "37 hours" censorship claim, found a documented instance of a White House official emailing Twitter (not Meta) on January 23, 2021, asking to have a different Kennedy post removed "ASAP"; that post was never taken down and stayed live, and no public record of a comparable White House communication to Meta specifically ordering the Instagram removal has surfaced. Kennedy's account had about 770,000 followers when it was reinstated in June 2023, a figure that does not match the "million followers" he says he lost in the removal. PolitiFact rated Kennedy's broader claim that Biden used federal power to censor a political opponent as False, noting Kennedy was not yet a declared political rival when the 2021 events occurred. The core fact that Instagram removed Kennedy's account in 2021 for COVID/vaccine misinformation is well established, but the specific causal claim of a direct White House order to Meta on a 37-hour timeline is not supported by the public record.

What the evidence shows: The 12,000 figure traces to a Times of London investigation (published April 2025, based on freedom-of-information requests to UK police forces) that found roughly 12,183 arrests in 2023, about 33 per day, under Section 127 of the Communications Act 2003 and Section 1 of the Malicious Communications Act 1988. Those laws cover grossly offensive, indecent, or menacing electronic communications sent by any medium, not only social media, so the total includes emails and messaging apps as well as posts. The figure was cited in UK parliamentary debate and widely reported, but it is a journalistic compilation from FOI data rather than a centrally published government statistic, and not all police forces supplied data. It also reflects arrests, not convictions: fewer than 1,119 people, under 10% of 2023 arrests, were sentenced that year, with cases commonly dropped for evidential difficulties or lack of victim support. Independent reporting confirms the underlying pattern of UK arrests over social media posts is real and has drawn international attention, including the high-profile September 2025 arrest, and subsequent no-charge decision, for writer Graham Linehan over posts on X. Rogan's framing, that 12,000 people were arrested specifically for social media posts within the most recent single year, compresses a broader 2023 statistic covering multiple offense categories and communication types into a narrower and more recent-sounding claim; the underlying number is grounded in real investigative reporting, but the precise scope and recency as stated on the podcast is imprecise.

What the evidence shows: Kennedy attributes to Ozempic a US-versus-UK price comparison (roughly $1,300+ in the US versus roughly $88-92 in the UK) that, per NPR's coverage of September 2024 Senate testimony from Novo Nordisk's CEO, was reported as Wegovy's pricing, not Ozempic's. NPR's report has Senator Sanders opening the hearing by citing Wegovy at $92/month in the UK versus more than $1,300 in the US, the same figures Kennedy applies to Ozempic. Ozempic and Wegovy are both Novo Nordisk semaglutide products but are dosed and priced differently, so applying one drug's figures to the other overstates the precision of the comparison. The broader point that US list prices for this drug class run far higher than in other wealthy countries is directionally consistent with the source, but the specific $1,350-versus-$88 figures as stated for Ozempic are not supported by the cited source.

What the evidence shows: TrumpRx.gov, a federal drug-discount platform, launched in February 2026 and lists reduced cash prices for dozens of brand-name drugs, including fertility medications used in IVF. The White House's own launch announcement states that fertility drug prices on the platform fall as low as $168 per pen (down from higher list prices) and that patients paying out of pocket for fertility treatment stand to save an average of $2,000+ per cycle. No White House release or other allowlisted source locates a specific IVF drug that dropped from $4,000 to about $600 as Kennedy described. The documented per-cycle savings figure ($2,000+) is lower than the roughly $3,400 reduction Kennedy implies, and the exact dollar amounts he cited do not match any publicly confirmed TrumpRx price. The broader claim that TrumpRx substantially cut IVF/fertility drug costs is supported by the White House's own figures, while Kennedy's specific $4,000-to-$600 figures are unconfirmed and likely an approximate, anecdotal recollection rather than a documented list price. Sources consulted: whitehouse.gov coverage of the February 2026 TrumpRx launch. A CBS News report on TrumpRx's published price list (Gonal-F at $168, an 83% discount) was reviewed but is not on this project's evidence allowlist and was excluded as a citation.

What the evidence shows: CDC's National Center for Health Statistics reports the US total fertility rate was approximately 1.62 births per woman in 2023 and 1.599 in 2024 (NCHS Data Brief No. 535, July 2025), the lowest level on record and well below the 2.1 replacement rate. The general fertility rate (births per 1,000 women aged 15-44) fell by only 1% from 2023 to 2024, to 53.8, which CDC's own summary describes as a modest year-over-year change rather than a dramatic single-year drop. No CDC data series shows the rate at 1.75 in the current period; a total fertility rate near 1.7 last occurred around the mid-1970s, decades before the timeframe Kennedy references. The 1.75 figure is inconsistent with official current data, which puts the rate lower, near 1.6, and describes a gradual multi-decade decline rather than a sudden one-year plunge.

What the evidence shows: In late 2023, under the Biden administration, the FDA reclassified more than a dozen popular peptides (including BPC-157 and others, reportedly around 19 total) as Category 2 bulk drug substances, a designation that bars compounding pharmacies from preparing them. Reporting from STAT and PolitiFact confirms the move was explicitly framed as safety-motivated, citing concerns such as lack of clinical trial data and unproven safety in humans, contradicting the claim that no safety rationale existed. Critics, including compounding advocates, have disputed whether the specific statutory 'safety signal' threshold was clearly met for every peptide on the list, but no reporting found supports the claim that the reclassification was done without any stated safety basis. As HHS Secretary, Kennedy, who has said he personally uses peptides, has pushed to reverse the restriction and make peptides more accessible again. Restricting compounded access has been linked by industry advocates and some reporting to a gray/black market for imported peptides sold with 'research use only' disclaimers, though this dynamic is not established as a direct causal FDA policy failure by independent reporting.

What the evidence shows: Kennedy characterizes marijuana and psychedelic legalization as historically an exclusively left-wing cause prior to Rick Perry's recent advocacy. Perry, as former Texas governor, did help drive the state's 2025 investment in ibogaine research and has spoken publicly about the drug's potential to treat veterans' PTSD and addiction, so the claim about Perry's own role is accurate. But the framing that drug-policy liberalization was purely left-wing before Perry omits a long-standing libertarian-right tradition: former Republican Congressman Ron Paul of Texas campaigned for full drug legalization, including marijuana and heroin, from the 1980s through his 2012 presidential run, and in 2011 co-sponsored federal legislation with Democratic Rep. Barney Frank to end federal marijuana prohibition and let states regulate it. This Republican strand of drug-policy reform, grounded in libertarian and states'-rights arguments rather than left-wing politics, predates Perry's ibogaine advocacy by more than a decade. The broader assertion that Republican support for legalization is new or exclusively associated with the left is not supported by the historical record.

What the evidence shows: Glyphosate use on U.S. corn rose sharply after glyphosate-tolerant ("Roundup Ready") corn was commercialized in 1996, and by the 2010s a large majority of U.S. corn acreage was planted with herbicide-tolerant varieties and treated with glyphosate at some point in the growing cycle, per peer-reviewed pesticide-use tracking. However, published USDA and peer-reviewed data do not report a 97% figure specifically for corn; herbicide-tolerant corn adoption and glyphosate-treated acreage have more commonly been cited in the 85-92% range depending on year and metric. The claim that corn "can't be produced without" glyphosate is inaccurate: non-genetically-engineered corn hybrids, conventional corn grown with alternative herbicides or mechanical weed control, and certified organic corn (which by definition excludes glyphosate) are all commercially produced in the United States. The core claim that glyphosate use in corn is widespread and increased substantially since the mid-1990s is well supported, but the specific 97% figure and the assertion of agronomic necessity are overstated.

What the evidence shows: Kennedy claimed that 99% of US glyphosate comes from China and that the Pentagon called this an "extreme national security vulnerability." China is a major supplier of the glyphosate imported into the US, but Bayer/Monsanto remains the sole domestic US glyphosate manufacturer and accounts for roughly 60% of glyphosate sold in the United States, meaning imports from all countries combined (not exclusively China) make up a minority share of US supply, not 99%. President Trump's February 2026 executive order (EO 14387), which invoked the Defense Production Act over glyphosate and elemental phosphorus supply chains, cites national-security and food-security risks from having only a single domestic producer, but the order's own text does not name China, does not cite a 99% figure, and contains no quoted Pentagon statement; it directs the Secretary of Agriculture to consult with the Secretary of War (the renamed Department of Defense), not a Pentagon pronouncement. The underlying concern, that the US relies on Chinese-sourced glyphosate and precursor inputs, reflects a real and documented supply-chain vulnerability, but the specific 99% statistic and the "Pentagon said" attribution are not substantiated by the primary government documents reviewed.

What the evidence shows: Peer-reviewed research on glyphosate's effects on soil microbial communities is mixed and inconsistent, not a settled finding that it "destroys" soil life: outcomes vary with soil type, application rate, and detection method, and a large multi-site field study (Maryland and Mississippi, 768 samples over two years) found no detectable change in overall soil microbial community composition, no pathogen enrichment, and no yield loss in glyphosate-treated corn and soybean systems relative to untreated plots. Separately, USDA data show glyphosate-tolerant soybean adoption is strongly associated with a shift to conservation and no-till farming (86% of glyphosate-tolerant soybean acreage under conservation tillage by 2006, versus 36% of conventional acreage), a practice widely credited in the agronomic literature with reducing soil erosion and improving water infiltration rather than causing the runoff and farmland destruction described. The claim that glyphosate broadly and definitively "destroys" the soil and drives farm-destroying runoff overstates a contested evidence base in which the dominant tillage-reduction pathway associated with glyphosate use is generally credited with erosion and infiltration benefits rather than harms.

What the evidence shows: Public reporting confirms the general scale of Bayer's Roundup litigation but not the precise figures Kennedy cites. Reuters reported in February 2026 that Bayer had previously paid about $10 billion (not $11 billion) to settle most Roundup lawsuits pending as of 2020, and that it proposed a new class-action settlement worth up to $7.25 billion (not $7.6 billion), which as of a June 30, 2026 hearing-date update was still awaiting court approval (an August 19 hearing) rather than "about to be signed." The 65,000-case figure is accurate: Reuters reports approximately 65,000 pending state and federal Roundup claims (AP separately cites about 200,000 total claims ever filed against Bayer, a different, larger cumulative figure). Bayer has already stopped selling glyphosate-based Roundup for the U.S. residential lawn-and-garden market and has said it "might have to consider" pulling glyphosate from U.S. agricultural markets if litigation continues, but this is a conditional statement rather than the flat declaration "we're getting out of this business" Kennedy attributes to the company. A June 2026 Supreme Court ruling in Bayer's favor has since reduced the near-term litigation pressure that would drive such an exit.

What the evidence shows: Kennedy's figure is corroborated by federal government sources: HHS, USDA, and EPA announced a joint investment of over $1 billion to accelerate farm modernization and long-term food supply security, an effort tied to the Make America Healthy Again Commission established by executive order in February 2025, predating this January 2026 podcast appearance. A subsequent executive order signed June 25, 2026, titled Advancing Regenerative Agriculture and Strengthening American Farm Resilience, formalized directives for agencies to research and develop technologies that reduce reliance on conventional chemical crop protection tools and improve chemical efficiency to reduce overall use. The $1 billion figure is real and officially documented, but it was announced as a broad farm-innovation and food-security investment rather than a sum specifically earmarked in dollar terms for chemical-reduction technology; the explicit federal directive tying research funding to reducing agricultural chemical use was formalized several months after this claim was made. Kennedy's characterization is therefore substantively accurate on the dollar figure and general thrust of administration policy, though it compresses a broader, multi-agency initiative into a single billion-dollar line item for chemical reduction specifically.

What the evidence shows: The specific dollar figures Kennedy cites for an unnamed South Texas onion farm ($1,500/acre dropping to $300/acre) cannot be independently verified; the operation, its acreage, and its exact costs are not named or documented in any public dataset, and the figures resemble the kind of case-study numbers laser-weeding vendors such as Carbon Robotics circulate in marketing materials. Independent, peer-reviewed research does support the underlying premise that laser weeding is a real, effective alternative to herbicides: a 2025 Cornell/Rutgers field study published in Pest Management Science found laser treatment reduced weed biomass by 97% or more in peas, beets and spinach, performed as well as or better than conventional herbicides including S-metolachlor, and caused far less crop damage (under 1% stunting versus a 34% head-weight reduction from S-metolachlor). That same peer-reviewed study cites a separate Western Growers industry case study reporting weeding costs falling from about $900/acre to roughly $269/acre on a large operation, a similar magnitude of savings to Kennedy's claim but with different numbers and a different farm. Laser-weeding units cost roughly $1-1.5 million, consistent with the payback-period framing Kennedy used elsewhere in the same conversation. Overall status: the general claim that laser weeding can substantially cut per-acre weed-control costs versus glyphosate and hand labor is well-supported by independent research, but Kennedy's specific onion-farmer figures are an unverified anecdote that could not be confirmed against any allowlisted primary source.

What the evidence shows: The mass Roundup litigation against Monsanto/Bayer centered on the International Agency for Research on Cancer's (IARC) March 2015 classification of glyphosate as a Group 2A "probable human carcinogen," a determination based specifically on human evidence linking occupational glyphosate exposure to non-Hodgkin lymphoma, matching Kennedy's claim. Separately, a 2020 human study of 97 biopsy-confirmed NAFLD patients at UC San Diego found that urinary glyphosate and AMPA excretion were significantly higher in patients with steatohepatitis (NASH, p=0.008-0.022) and correlated with fibrosis severity (p=0.001-0.046), showing a genuine but modest statistical association. That study had no non-NAFLD control group and could not establish causation, and no major regulator treats glyphosate-NAFLD as an established causal link: the EPA's current position is that glyphosate is "not likely to be carcinogenic to humans," and its glyphosate materials do not list liver disease as a recognized human health endpoint of concern. Kennedy's statement is therefore roughly accurate on both points: the litigation was indeed built on non-Hodgkin lymphoma evidence, and a documented but preliminary, non-causal association with liver disease exists, though calling the liver finding a settled "scientific association" somewhat overstates evidence drawn from a single uncontrolled human cohort.

What the evidence shows: Kennedy claimed glyphosate use as a pre-harvest wheat desiccant disrupts the "plant" component of the gut microbiome and is a major driver of celiac disease and gluten sensitivity. Established medical guidance holds that celiac disease is primarily a genetic autoimmune condition, occurring in people carrying HLA-DQ2 or HLA-DQ8 gene variants whose immune system reacts to gluten, damaging the small intestine; environmental and other factors are studied as possible contributors but glyphosate is not established as a cause. A 2020 peer-reviewed critical review in Frontiers in Microbiology examined the glyphosate-dysbiosis-wheat-sensitivity hypothesis in detail and found that while glyphosate residues could theoretically alter gut bacteria, the existing research on this link "suffers from numerous methodological weaknesses" that make it "impossible to draw any definitive conclusions" about glyphosate's effect on human health through the microbiome. No robust clinical or epidemiological evidence has demonstrated that glyphosate causes or meaningfully contributes to celiac disease in humans. The claim remains an unproven hypothesis rather than an evidence-based causal mechanism, and Kennedy himself acknowledges elsewhere in the same discussion that clear scientific evidence for the claim is lacking.

What the evidence shows: Rogan relays a single secondhand anecdote that a former guest's "brain fog" cleared after stopping aspartame consumption, presenting an individual, unverified experience as though it demonstrates a causal effect. Major regulatory and scientific bodies have extensively reviewed aspartame's health effects: a July 2023 joint assessment by the WHO/IARC and the FAO/WHO Joint Expert Committee on Food Additives (JECFA) classified aspartame as "possibly carcinogenic" (IARC Group 2B) based on "limited evidence," while JECFA reaffirmed the existing acceptable daily intake of 40 mg/kg body weight, finding no other safety concerns at typical consumption levels. The U.S. FDA has stated it does not have safety concerns about aspartame under approved conditions of use and noted shortcomings in the studies IARC relied on. Neither body's review identifies reliable evidence that aspartame causes "brain fog" or that discontinuing it reverses cognitive symptoms. The claim as stated is an unsubstantiated personal anecdote rather than an evidence-based causal finding.

What the evidence shows: According to the FDA's own regulatory history, an independent Public Board of Inquiry (PBOI) of scientists, not a named FDA commissioner, revoked aspartame's 1974 approval in October 1980 over unresolved questions about brain tumor risk. In July 1981, the FDA's Commissioner overturned the PBOI's decision and allowed aspartame's marketing in dry foods. Historical accounts (a peer-reviewed Virginia Tech Undergraduate Historical Review article, drawing on GAO reports and contemporaneous press coverage) identify that commissioner as Arthur Hull Hayes Jr., who was appointed in 1981 after Donald Rumsfeld, then CEO of aspartame's manufacturer G.D. Searle and a member of Reagan's transition team, pushed to replace the sitting commissioner; Hayes then approved aspartame months later over the objections of FDA scientists. Kennedy's claim is directionally accurate that an FDA ruling against aspartame was reversed following Reagan-era personnel changes tied to Searle's leadership, but the quote garbles the names and mechanism: it is unclear whether he means David Kennedy (an actual former FDA commissioner from 1969-1971, unrelated to him, who had no role in the aspartame decisions) or David Kessler (who did not become FDA commissioner until 1990, under George H.W. Bush, and also had no role in the 1980-1981 aspartame decisions). Neither man banned aspartame, and Rumsfeld did not personally overrule an FDA head; rather, he helped install Hayes as commissioner, and Hayes then overturned the PBOI's independent scientific ruling. Status: directionally accurate on the Rumsfeld/Searle conflict-of-interest narrative, but the names and mechanism as stated are garbled.

What the evidence shows: Kennedy's two figures echo a real, ongoing dispute between the Trump administration and independent analysts over how to characterize ICE arrest data. DHS's own January 20, 2026 statement said '70% of ICE arrests are of criminal illegal aliens who have been convicted or charged with a crime in the U.S.,' and border czar Tom Homan has cited the same '60% to 70%' range; Homeland Security Secretary Kristi Noem, in a January 18, 2026 CBS interview, used the 70% figure but described it specifically as violent crimes, a framing CBS immediately disputed. Independent analysis by FactCheck.org (using Deportation Data Project records covering Trump's first year) found that arrestees with either a conviction or a pending charge, of any kind, made up about 66%, which the administration rounds to 70%; on a snapshot basis, PolitiFact found roughly half of the currently detained population and about 64% of everyone detained since January 2025 had a conviction or pending charge. On the violent-crime-specific figure, an internal DHS document reported by CBS News shows about 13.9% of arrestees over Trump's first year had charges or convictions for violent offenses (homicide, robbery, sexual assault, kidnapping, arson), closely matching the '14%' Kennedy attributes to Democrats; separate analyses by the Cato Institute (about 5%) and the New York Times (about 7%) put the violent-conviction share even lower. So Kennedy's 14% figure is roughly consistent with independent reporting on violent crime specifically, but his 70% figure conflates two different things: it is a defensible round number for 'any criminal charge or conviction' (matching DHS's own framing and FactCheck.org's 66%), not for violent crime, despite Kennedy (following Noem's ambiguous wording) placing it in contrast with a violent-crime statistic as though both describe the same category. Independent tracking also shows the share of arrestees with no criminal record or charges at all has been rising through the year, from about 22% in Trump's first three months to about 43% by January 2026, undercutting the administration's 'worst of the worst' framing.

What the evidence shows: Kennedy cited an unnamed chart to claim more American citizens were arrested during immigration enforcement under Obama than under Trump, but did not identify the chart's source, methodology, or time frame, and no chart or dataset specifically measuring U.S. citizen arrests by administration has been located. Published government and independent data instead track total deportations and the criminal-record status of non-citizens arrested by ICE, not arrests of American citizens. On the closely related and often-conflated claim that Obama removed more people overall than Trump, PolitiFact rated a similar comparison "True": DHS data show roughly 3.2 million removals and returns in Obama's first term and 2.1 million in his second, versus about 2 million under Trump's first term. Separately, FactCheck.org and NPR reported that under Trump's second term, the share of ICE arrestees with no U.S. criminal record rose sharply, from about 22% early in the term to over 40% by January, a trend that runs counter to Kennedy's framing that non-criminals were targeted more heavily under Obama. No dataset was found that isolates "American citizens" arrested by immigration authorities under either administration, so the specific claim as stated remains unsupported by available evidence.