After Dobbs, a Computer Scientist Targets Contraceptive Misinformation Online

After the Supreme Court's 2022 Dobbs decision, millions of Americans saw their access to reproductive healthcare shrink almost overnight. Many did what people do when they need answers and don't know where else to turn: they went online. They scrolled through TikTok. They browsed Reddit. They watched YouTube videos, looking for guidance on birth control. What they often found was not carefully vetted medical information, but viral "horror stories" and misleading claims, amplified by algorithms designed to keep them clicking rather than to keep them safe.

With a new two-year grant from the Society of Family Planning's Contraceptive Misinformation and Disinformation initiative, Stony Brook University computer scientist Ritwik Banerjee is bringing a different kind of expertise to challenge this growing problem. An assistant professor of research in the Department of Computer Science, Banerjee is the only member of his grant's cohort coming from computer science rather than public health or medicine-positioning him to study not just what people are saying about birth control online, but how they are framing these issues in community-specific ways to help them spread across recommendation systems.

Banerjee's project will look at more than a million social media posts about contraceptive health, instead of reviewing a small sample by hand. Using transformer-based language models, his team will identify recurring themes, frames, and rhetorical patterns linked to misinformation. As part of the project, the researchers will design AI agents to simulate how different users might encounter this content online, including teenagers looking up side effects or adults living in states with new restrictions.

Additionally, the project will trace how information moves across platforms, showing how a single TikTok video can spread into Reddit discussions and eventually distort medical facts across the broader online ecosystem.

"Computational tools for language analysis let us see the ecosystem, not just a small sample," Banerjee explains.

For him, that ecosystem is deeply tied to real people and real harms. Drawing on personal and academic experience with cross-cultural health narratives, Banerjee has seen how health-related stigma and misinformation can fall hardest on communities that are already underserved, even as the channels for spreading that stigma shift from word-of-mouth to algorithmically curated feeds. He also points to a "glaring research gap": while vaccine misinformation has been mapped in detail, contraceptives have too often been overlooked as a focus of rigorous, data-driven study.

By the end of the two-year grant, Banerjee wants the project's impact to be tangible. The team plans to release open-source natural language processing plugins that health departments can use to monitor emerging myths, along with practical API guidelines that platforms could adopt to de-amplify harmful content and policy benchmarks for holding recommendation algorithms more accountable in reproductive health.

"I want a world where a health department with limited resources can use what we build; these tools should not be just for universities or big tech companies," he says.

For Banerjee, the goal is not only to better understand contraceptive misinformation online, but to give public health practitioners new tools to recognize it, respond to it, and reduce its harm before it spreads.