About me
I am a Research Scientist at FAIR Paris (Fundamental AI Research lab at Meta).
My research focuses on content protection and safety in machine learning, with a particular interest in watermarking generative models. My goal is to make content generated by AI traceable and identifiable. I notably developed:
- the audio watermarking algorithm used in public research demos from FAIR, such as Audiobox and Seamless, serving 100k users daily (joint work with Hady Elsahar);
- the image watermarking algorithm used by Imagine;
- Stable Signature, a method to watermark generative models even if they are open-sourced.
I did my PhD at FAIR and INRIA Rennes, advised by Matthijs Douze, Hervé Jégou and Teddy Furon. Prior to my PhD I studied at École polytechnique, majoring in computer science and mathematics for visual computing. I also hold a Master’s degree from École des Ponts and from Paris-Saclay University where I studied Mathematics for vision and learning. Prior to my PhD, I studied at École polytechnique, majoring in computer science and mathematics for visual computing. I also hold a Master’s degree from École des Ponts and from Paris-Saclay University.
News
Oct 29, 2024 | Our paper on Radioactivity of LLM Watermarking is accepted at NeurIPS as a spotlight. |
---|---|
Jun 18, 2024 | AudioSeal is featured on MIT Tech Review and on Meta AI blog! |
Jan 30, 2024 | AudioSeal, a state-of-the-art localized watermarking method, that can be used to proactively detect voice cloning. |
Dec 8, 2023 | Our paper Three Bricks to Consolidate Watermarks for LLMs was awarded the best student paper award at WIFS 2023! |
Dec 1, 2023 | Our team releases Seamless Streaming and Expressive, the best speech/text translation systems in the world, and the first to include watermarking. |
Sep 18, 2023 | Attending ELLIS Summer School on Large-Scale AI. |
Sep 17, 2023 | Our work on text watermarking Three Bricks to Consolidate Watermarks for LLMs is accepted at WIFS! |
Jul 14, 2023 | Stable Signature is accepted at ICCV 2023 in Paris. TL;DR: how to watermark your diffusion models and all images they produce! |
Jun 16, 2023 | I took part in the GenAI pitch as a juror, at VivaTech 2023. Replay is accessible here. |
Apr 14, 2023 | Our team at Meta AI releases DINOv2, a family of foundation models producing high-performance visual features. |