Nicolas Flammarion

TML, IC, EPFL. Profile page

Nicolas Flammarion is mapped on the EPFL MDH Research Landscape. 7 of their works are recorded here.

Works recorded here

• Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks
  publication, 2025, adjacent MDH relevance
• Competition Report: Finding Universal Jailbreak Backdoors in Aligned LLMs
  publication, 2024, adjacent MDH relevance
• Does Refusal Training in LLMs Generalize to the Past Tense?
  publication, 2025, adjacent MDH relevance
• HalluHard: A Hard Multi-Turn Hallucination Benchmark
  publication, 2026, adjacent MDH relevance
• JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models
  publication, 2024, adjacent MDH relevance
• On the Adversarial Robustness of Discrete Image Tokenizers
  publication, 2026, adjacent MDH relevance
• OS-Harm: A Benchmark for Measuring Safety of Computer Use Agents
  publication, 2025, adjacent MDH relevance

Open this actor on the interactive map

Part of the EPFL MDH Research Landscape, a map of research at EPFL on misinformation, disinformation and hate speech. The whole dataset is at data.json and in one readable file at mdh-landscape.md.