Baby-Sitters Flub
HHS exposed in court for citing fake AI-generated studies while attempting to weaken the Teen Pregnancy Prevention Program
By Elizabeth "Bitsy" Skerry
The U.S. Department of Health and Human Services (HHS) was recently admonished in court for citing nonexistent studies, presumably because the agency relied on AI to generate citations. HHS used the non-existent studies to justify sabotaging the highly successful Teen Pregnancy Prevention (TPP) Program.
For context, in June, HHS tried to change the statutorily mandated TPP Program, which provides grants for evidence-based programs to prevent teen pregnancy as well as sexually transmitted infections and research to develop new programs. Instead, HHS wanted to award grants exclusively to recipients who promote abstinence-only pregnancy prevention. That new policy was challenged in court, and the judge put it on hold while the case proceeds.
In his opinion, the judge noted the agency’s reliance on “hallucinated” studies, i.e., ones that simply do not exist. He wrote, “On the topic of body literacy, the notices (remarkably) reference public health studies that appear either not to exist or not to support the propositions for which they are cited—a hallmark of AI-generated citations.”
HHS’s use of fake AI-generated studies is, at best, sloppy and, at worst, dishonest. This example shows why agencies, if they choose to use AI, must have a human verification system in place to ensure AI systems are not making up citations to sycophantically justify the agency’s position and desired objective. As Public Citizen has proposed in our guardrails for the use of AI in the federal regulatory process, agency decisions, and the justifications for those decisions, should never be fully automated or rubber stamped.
Given the real life consequences of this rulemaking snafu, it serves as an unfortunate example of how pressing a need there is to see these guardrails enacted as soon as possible.