Ilia Shumailov

Soft Instruction De-escalation Defense

The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against LLM Jailbreaks and Prompt Injections

Exploiting Reasoning Patterns in Language Models for Indirect Targeted Poisoning

Exploring the limits of strong membership inference attacks on large language models

Measuring memorization in RLHF for code completion

Interpreting the Repeated Token Phenomenon in Large Language Models

Breach By A Thousand Leaks: Unsafe Information Leakage in 'Safe' AI Responses

Hardware and Software Platform Inference

Stealing User Prompts from Mixture-of-Experts Models