AI Provides Evidence-Based Guidance on Acetaminophen Use During Pregnancy

Artificial intelligence demonstrated an ability to prioritize evidence-based research over sensationalized content when queried about the safety of acetaminophen use during pregnancy, according to data presented at the ACOG Annual Clinical & Scientific Meeting on May 2, 2026.

The findings, shared by medical professionals at the conference, suggest that large language models may be capable of navigating complex medical contradictions without defaulting to high-engagement or inflammatory responses.

“I was unexpectedly relieved by the information that ChatGPT provided to our pregnant test subject,” Erik Holder, MD, MPH, an OB/GYN specialist at The University of Texas Health Science Center at San Antonio

Dr. Holder noted that his primary concern was that the AI might provide a click-bait answer or emphasize conflicting data simply to maintain user engagement, rather than providing a balanced, evidence-based perspective.

The Acetaminophen Debate in Pregnancy

The ability of AI to handle queries regarding acetaminophen is particularly significant due to the ongoing clinical dialogue surrounding its use. Acetaminophen is widely regarded as the first-line analgesic and antipyretic for pregnant individuals due to its long history of use and perceived safety profile compared to other pain medications.

However, the conflict of information referenced by Dr. Holder stems from various observational studies. Some research has suggested potential associations between prolonged acetaminophen use during pregnancy and an increased risk of neurodevelopmental or behavioral issues in children, such as attention-deficit/hyperactivity disorder (ADHD) or autism spectrum disorder (ASD).

Medical experts frequently caution that these observational findings are often confounded by the underlying reasons for medication use. For example, the maternal infection or high fever that prompted the use of acetaminophen may be the actual driver of the developmental outcomes, rather than the medication itself.

Because of this tension between broad clinical guidelines and specific observational reports, patients often encounter contradictory information when searching for health advice online.

AI and Medical Information Accuracy

The presentation at the ACOG meeting highlights a critical intersection between generative AI and patient education. The risk of AI producing hallucinations—factually incorrect but confident-sounding statements—has been a primary concern for the medical community.

When AI prioritizes evidence-based research, it helps mitigate the risk of patients making healthcare decisions based on anecdotal evidence or misinterpreted data. By providing a nuanced view of the research, the tool can potentially reduce patient anxiety while still encouraging consultation with a healthcare provider.

The use of a pregnant test subject in this instance allowed researchers to observe how the AI interacts with a user in a high-stakes health scenario, where the balance between risk and benefit is a central concern for both the patient and the clinician.

Clinical Implications

While the results from the ACOG meeting are encouraging, the integration of AI into patient health journeys remains a subject of rigorous scrutiny. The ability of a model to provide a correct answer in a specific test case does not eliminate the need for professional medical oversight.

Clinicians continue to emphasize that medication use during pregnancy should be individualized. The general medical consensus remains focused on using the lowest effective dose for the shortest possible duration to manage symptoms.

The data suggests that as AI models are refined, they may become more effective tools for synthesizing medical literature, provided they are designed to resist the incentives of engagement-driven content and instead adhere to established clinical evidence.