Research project 2026 Completed
Research on inference attacks and defences for systems that use large language models to identify phishing emails.
Research question
Frederik Sherman and Yoni Birman investigate inference-time attacks against language-model systems used to identify phishing emails, alongside defences for those systems.
The work was presented at the USENIX Security 2026 poster session and in a paper at HumSec 2026, an ESORICS workshop. The linked publications below bring these outputs together with their original sources.
HumSec 2026 · ESORICS workshop
Frederik Sherman and Yoni Birman study inference attacks and defenses for LLM-powered phishing email detection. The paper is available through the HumSec 2026 workshop.
USENIX Security 2026 · Poster session
Frederik Sherman and Yoni Birman explore attacks and defenses involving LLM-powered phishing detection, presented at the USENIX Security 2026 poster session.
Explore the people and research areas connected to this project.
Discuss collaboration ↗
