Manuel Tonneau
PhD candidate in Social Data Science
Oxford Internet Institute
University of Oxford
About
I am an incoming postdoctoral researcher at the Hasso Plattner Institute of the University of Potsdam. I am currently finishing my Ph.D. in Social Data Science at the University of Oxford, where I am supervised by Scott Hale and Ralph Schroeder. I also consult for the World Bank and am affiliated with the Weizenbaum Institute. Prior to my Ph.D., I trained at the intersection of the social sciences and mathematics before specializing in statistics and economics at ENSAE Paris and the Humboldt University of Berlin, followed by applied roles across government and tech.
Research
I study the societal impact of AI, working at the intersection of machine learning, natural language processing, and computational social science. A central focus of my work is evaluation: building rigorous tests of how AI systems behave in the real world, and questioning the data and methods used to measure that behavior. My doctoral research turned this lens on how AI systems moderate online speech, showing that social media platforms' one-size-fits-all approach to hate speech fails to deliver the shared meaning, equitable resources, and effective protection it promises users across geographies. For this work, I received an Outstanding Paper Award at ACL.
All my publications can be found on my Google Scholar. My research has appeared in PNAS Nexus, ACL, ICWSM and FAccT and has been covered by Al Jazeera, Tagesspiegel, the MIT Technology Review and more.
My research is supported by a Global Merit Award from the Shirley Scholars Fund and by the Oxford Internet Institute Research Programme on AI, Government and Politics.
News
| Jun '26 | 💬 I was interviewed by the Daily Mirror on how social media platforms amplify and accelerate offline tensions in the aftermath of the Henry Nowak murder. |
| Apr '26 | 📜 New preprint on the closing of the Perspective API and its lessons for building transparent, reproducible toxicity measurement infrastructure. |
| Mar '26 | 📜 Paper accepted at FAccT: "Language Disparities in Moderation Workforce Allocation by Social Media Platforms" (preprint here) |
| Jan '26 | 📜 New preprint on the validity of demographic cueing of LLMs |
| Dec '25 | 📜 New paper in PNAS Nexus: "Can social media reliably estimate unemployment?" |
| Dec '25 | 🏆 Awarded a Dieter Schwarz Foundation–OII grant for research on AI, Government and Policy |
| Oct '25 | 💬 I was interviewed on the Somewhere On Earth podcast about our preprint on language disparities in moderation workforce allocation. |
| Jul '25 | 🏆 HateDay got an Outstanding Paper Award at ACL 2025! |
| May '25 | 📜 Two papers accepted at ACL 2025: "HateDay: Insights from a Global Hate Speech Dataset Representative of a Day on Twitter" (Main) and "When Claims Evolve: Evaluating and Enhancing the Robustness of Embedding Models Against Misinformation Edits" (Findings) |
| May '25 | 💬 Gave a talk at the MilaNLP lab of Bocconi University (photo) |
| Apr '25 | 💬 Gave a talk at the Hertie School Data Science Lab (blog post, video) |
| Feb '25 | 💬 Gave a virtual talk on HateDay at the Stanford Security Lunch Seminar (photo) |
| Dec '24 | 💬 Gave feedback on the EU Digital Services Act with OII colleagues. Read our blog post. |
| Nov '24 | 🤗 Our hate speech supersets, combining all hate speech corpora in 8 languages, have reached 2K downloads on Hugging Face! |
| Oct '24 | 🇩🇪 Started a fellowship at the Weizenbaum Institute in Berlin, Germany, following an invitation from Elizaveta Kuznetsova |
| Summer '24 | 💬 Presented our work on cultural bias in hate speech datasets (link) at WOAH and C3NLP |
| May '24 | 📜 Paper accepted at ACL 2024: "NaijaHate: Evaluating Hate Speech Detection on Nigerian Twitter Using Representative Data" (preprint here) |
| Apr '24 | 📜 Paper accepted at the Workshop on Online Abuse and Harms (WOAH), co-located at NAACL 2024: "From Languages to Geographies: Towards Evaluating Cultural Bias in Hate Speech Datasets" (link) |
| Mar '24 | 📜 Paper accepted at ICWSM 2024: "280 Characters to Employment: Using Twitter to Quantify Job Vacancies" |
| Oct '23 | 💬 Gave a talk at the Weizenbaum Institute in Berlin, Germany |
| Jul '23 | 💬 Presented ongoing work on hate speech on Nigerian Twitter at IC2S2 (photo) |
| Mar '23 | 📜 Paper accepted at ICWSM 2023: "Large-Scale Demographic Inference of Social Media Users in a Low-Resource Scenario" (link) |
