Unraveling the Errors, Hallucinations, and Questionable Science in a White House Health Initiative
- The “Make America Healthy Again” (MAHA) report, led by Health Secretary Robert F. Kennedy Jr., is riddled with citation errors, including nonexistent studies and AI-generated markers like “oaicite,” pointing to the use of tools like ChatGPT.
- Investigations by NOTUS and The Washington Post reveal dozens of inaccuracies, from broken links to misstated conclusions, casting doubt on the report’s credibility as a transformative health policy document.
- Despite official defenses labeling errors as mere “formatting issues,” experts and critics argue the reliance on AI and sloppy research undermines the report’s validity for policymaking, raising broader concerns about AI in government.
The “Make America Healthy Again” (MAHA) report, a flagship health initiative under the Trump administration and spearheaded by Health and Human Services Secretary Robert F. Kennedy Jr., promised to tackle the alarming decline in U.S. life expectancy with “gold-standard” science. Backed by over 500 citations, the report aimed to pinpoint root causes like environmental toxins, poor nutrition, and excessive screen time. Yet, beneath its ambitious rhetoric lies a troubling reality: the document appears to be a patchwork of errors, fabricated sources, and potential AI-generated content, calling into question its reliability and the broader implications of using artificial intelligence in government policy.
Investigations by NOTUS and The Washington Post have exposed a litany of issues within the MAHA report’s 522 citations. Dozens of references contain broken links, incorrect authors, wrong issue numbers, or misstated conclusions. More alarmingly, at least seven cited studies simply don’t exist, while 37 citations are repeated multiple times, suggesting a lack of thoroughness. Some URLs even carry the “oaicite” marker—a telltale sign of content sourced from OpenAI models like ChatGPT. This isn’t just a minor oversight; it’s a glaring red flag about the report’s foundation. AI tools, while powerful for summarizing research, are notorious for “hallucinations”—fabricating information that seems plausible but is entirely false. As Oren Etzioni, a professor emeritus at the University of Washington, bluntly put it, “Frankly, that’s shoddy work. We deserve better.”
Specific examples of these errors paint a damning picture. One citation referenced a nonexistent study titled “Overprescribing of Oral Corticosteroids for Children With Asthma” to support claims of overmedication, only to be quietly replaced with a real 2017 Pediatrics article after scrutiny. Another reference to a U.S. News & World Report story on children’s physical activity credited fictitious authors, later corrected to the actual writer, Kate Rix. At least 21 links in the report were dead, and even credible sources were sometimes misrepresented to bolster the report’s conclusions. These aren’t just clerical mistakes; they suggest a deeper reliance on unverified, potentially AI-generated content that undermines the report’s claim to scientific rigor.
The White House and Department of Health and Human Services have scrambled to downplay the controversy. Press Secretary Karoline Leavitt dismissed the errors as “formatting issues,” insisting the report remains a transformative piece backed by unrecognized “good science.” Spokesman Andrew Nixon echoed this sentiment, stating that while “minor citation and formatting errors have been corrected,” the substance of the MAHA report—a historic assessment of the chronic disease epidemic among children—remains unchanged. Updates to the online document on Thursday removed some “oaicite” markers and swapped out nonexistent sources, but critics argue these fixes are too little, too late. As Georges C. Benjamin of the American Public Health Association starkly warned, “This is not an evidence-based report… it should be junked at this point.”
Public and expert backlash has been swift and scathing. Senators Chris Van Hollen and Elizabeth Warren condemned the administration’s apparent use of AI-generated content in a policy document meant to protect American health, with Warren calling it “shameful” that parents must question the authenticity of White House science. Anand Parekh of the Bipartisan Policy Center labeled the episode a “cautionary tale” for AI in government, questioning whether inadequate staffing or oversight led to such sloppy work. Peter Lurie of the Center for Science in the Public Interest, a former FDA official, noted the hypocrisy of RFK Jr.—a vocal advocate for AI in healthcare—producing a report cloaked in scientific excellence yet reliant on questionable tech. Steven Piantadosi, a professor at UC Berkeley, further criticized modern AI’s lack of trustworthiness, explaining that it operates on statistical associations rather than grounded evidence or logical rigor.
RFK Jr.’s enthusiasm for AI isn’t new. He has long championed the “AI revolution,” touting its potential to streamline healthcare data and even highlighting an AI nurse prototype for rural areas during a congressional hearing. But this incident highlights the pitfalls of unchecked reliance on such tools. Legal filings and other policy documents, like a recent housing report by Andrew M. Cuomo, have similarly faced scrutiny for AI-generated errors, with attorneys sanctioned for citing nonexistent cases. The MAHA report’s issues aren’t isolated—they reflect a broader challenge of balancing AI’s efficiency with the need for accuracy and accountability in high-stakes contexts.
What does this mean for the future of health policy under this administration? The MAHA report was intended as a bold step to address a national crisis, yet its credibility is now in tatters. Beyond the immediate embarrassment, the episode raises critical questions about how emerging technologies are vetted before shaping decisions that affect millions. If AI is to play a role in government, as RFK Jr. envisions, robust checks and human oversight must be non-negotiable. For now, the report stands as a stark reminder that cutting-edge tools can’t replace the painstaking work of genuine research and validation.
As the dust settles, one thing is clear: the “Make America Healthy Again” initiative, meant to inspire confidence in federal health leadership, has instead fueled skepticism. Whether the administration can salvage its reputation—or the report’s intent—remains to be seen. But for Americans awaiting real solutions to a declining life expectancy, the stakes couldn’t be higher, and the demand for trustworthy science has never been more urgent. What are your thoughts on this mix of AI and policy? Should there be stricter guidelines for tech in government reports, or is this just a bumpy start to a necessary innovation?