Fourteen peer-reviewed papers, each one measured in the field, shipped as open infrastructure, or used in policy briefings.
Research that is used by frontline health workers, journalists, regulators, and the public. Our work spans information integrity, recommender systems, election integrity, decentralized platforms, AI for health, and AI education. Filter by topic and venue, or search by phrase.
Where we've written, signed, or been featured.
Who captures the value journalists create?
Newsrooms create the content. AI platforms control distribution. And they're claiming the revenue too. Notes for newsrooms on what they can take from how tech built distribution from scratch.
Newsrooms are navigating the unenviable challenge of sustaining revenues and audiences in an age of AI-generated content while maintaining the quality of their journalism. The recurring theme across marketplace history is that if distribution is controlled by a digital intermediary, there is always a risk that they may capture the value created by a publisher. When content is abundant and cheap due to LLMs, sensemaking becomes a scarce and valuable commodity.Read on akademie.dw.com
Using automation to fight misinformation, starting with a menstrual-health chatbot
Rest of World profiles SimPPL and Swapneel's framing of how automated tooling, beginning with the Sakhi maternal-health chatbot in Bengali and English, can support populations underserved by mainstream platforms.
Swapneel Mehta is the founder of the Simppl research collective, a group of students and professional programmers working on automated social media tools.Read on restofworld.org
BU professors earn NSF grant for data-science project fighting online misinformation
BU CDS announces a $550K NSF award to develop technology-based solutions for reducing online misinformation while preserving free speech. Swapneel is named on the project team alongside Prof. Ran Canetti and Prof. Nina Mazar.
The Boston University team includes Professor Ran Canetti and Professor Nina Mazar, with Swapneel Mehta as a postdoctoral researcher on the project.Read on bu.edu
Election Deepfakes: What To Do About Political Media That Doesn't Mean What You Think It Means
Joint comment co-authored by seven Integrity Institute members, including Swapneel, on how the U.S. Federal Election Commission should approach deepfake political advertising in the 2024 cycle.
Read on integrityinstitute.orgWhy AI May Make Integrity Jobs Harder
How generative AI shifts the workload for trust-and-safety teams. Co-authored with Sarah Amos, Rebecca Thein, and Soribel Feliz.
Read on integrityinstitute.orgComment on Meta's Approach to the Term “Shaheed”
Joint comment to the Meta Oversight Board on platform policy around the Arabic term “Shaheed” and content moderation across geographies.
Read on integrityinstitute.orgUnleashing the Potential of Generative AI in Integrity, Trust & Safety Work
On the opportunities, challenges, and concrete solutions for using generative AI in integrity and trust-and-safety teams.
Read on integrityinstitute.orgThe Ultimate Guide to Election Integrity, Part II
Podcast episode discussing the Integrity Institute's election-integrity guide. Companion document at wjngs.issuelab.org/resources/44004.
Read on integrityinstitute.orgThree projects deployed with partners.
Beyond the peer-reviewed track: software that ships to a UN mission, a public-radio newsroom, and platform research teams. Infrastructure, not papers.
Multimodal querying system for UN-mission analysts
RAG-based search over text and audio knowledge bases, so analysts in conflict regions can surface narratives from hours of radio data in minutes.
Axioma, newsroom audience analytics for WNYC
A Google Analytics plugin piloted with New York Public Radio over twelve months. 20M users, 136M lifecycle events, and a richer picture of who reads which authors and tags.
YouTube Data Collection and Coordinated Network Analysis
440K videos, 40M comments, 1.6K channels harvested via the YouTube Data API on Vertex AI and stored in BigQuery, used to study coordinated cross-platform link-sharing.
Investigations and CIB reports.
Coordinated-inauthentic-behavior reports we co-authored with partner organizations. Each entry links to the public-facing version on the partner's site.
- Q1 2024→Bangladesh political disinformation, with Tech Global Institute (1 of 2)
How Facebook has become a political battleground in Bangladesh. Fed Meta's public takedown announcement Q1 2024.
- Q1 2024→Bangladesh gendered disinformation, with Tech Global Institute (2 of 2)
From homophobia to assault. The gendered landscape of Bangladesh's political disinformation.
- 2023→Twitter pro-Russian bot networks, article-sharing analysis (1 of 2)
Unraveling coordinated behaviour: how user groups and individuals share articles, and how influential actors amplify them. Surfaced findings that prompted a Twitter investigation.
- 2023→Twitter pro-Russian bot networks, active-vs-deleted user graphs (2 of 2)
Network graphs over active and deleted accounts, tracing how information disseminates through coordinated structures.