Postdoc, CISPA Helmholtz Center for Information Security
2 papers at NeurIPS 2025
This work evaluates LLMs' utility and risk in a concurrent scenario and proposes a novel jailbreak attack against LLMs based on task concurrency.
