🧑‍🔬 Nicolas Papernot

🏛️ Affiliations

🎓University of TorontoToronto · Canada2019–2026 · 96 papersEmploymentcurrent🔬Vector InstituteToronto · Canada2019–2026 · 80 papersEmployment since 2019current🎓Pennsylvania State UniversityState College · United States2014–2019 · 20 papersEducation 2014–2016Education 2016–2018🏢GoogleMountain View · United States2018–2024 · 18 papers🎓University of CambridgeCambridge · United Kingdom2021–2022 · 2 papers🎓École Centrale de LyonÉcully · France2012–2016Education 2012–2016🎓Lycée Louis-le-GrandParis · France2010–2012Education 2010–2012

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 11 more institutions on single bylines are not listed.

🧭 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📅 Works by year

2020s

📄 Papers

The most cited first (OpenAlex and our library).

1. AI models collapse when trained on recursively generated data (2024 · 863 citations)
   The paper examines what happens to generative models once their own outputs make up much of the web data used to train their successors.
2. The Curse of Recursion: Training on Generated Data Makes Models Forget (2023 · 255 citations)
   Generative models are increasingly filling the web with text and images, so future models will likely be trained partly on their predecessors' output.

All works

5 works of theirs in our library, the most cited first; 2 with a page here, the others link their original until we summarise them.

1. AI models collapse when trained on recursively generated dataNatureIlia Shumailov, Zakhar Shumaylov, Yiren Zhao, et al.2024 · 863 citations
2. MixMatch: A Holistic Approach to Semi-Supervised Learning ↗arXiv (Cornell University) · no summary yetDavid Berthelot et al.2019 · 666 citations
3. The Curse of Recursion: Training on Generated Data Makes Models ForgetarXiv:2305.17493Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, et al.2023 · 255 citations
4. Tempered Sigmoid Activations for Deep Learning with Differential Privacy ↗Proceedings of the AAAI Conference on Artificial Intelligence · no summary yetNicolas Papernot et al.2021 · 104 citations
5. Language Models May Verbatim Complete Text They Were Not Explicitly Trained On ↗arXiv (Cornell University) · no summary yetKen Ziyu Liu et al.2025 · 3 citations

🤝 Co-authors here

5 co-authors with a page, the most papers together first.

• Ilia Shumailov ×2
• Ross Anderson ×2
• Yarin Gal ×2
• Yiren Zhao ×2
• Zakhar Shumaylov ×2

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

• Nature: Nature 2024
• Proceedings of the AAAI Conference on Artificial Intelligence

🔗 Cited by

Papers of our library cite their works 318 times (of the 149,439 papers whose bibliographies we read); 64 citing papers with a page here →.

Updated 2 November 2026

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/nicolas-papernot/