Ivan Evtimov

Large Reasoning Models Learn Better Alignment from Flawed Thinking

OpenApps: Simulating Environment Variations to Measure UI Agent Reliability

RL Is a Hammer and LLMs Are Nails: A Simple Reinforcement Learning Recipe for Strong Prompt Injection

AdvPrefix: An Objective for Nuanced LLM Jailbreaks

Automated Red Teaming with GOAT: the Generative Offensive Agent Tester

Persistent Pre-training Poisoning of LLMs

Gradient-based Jailbreak Images for Multimodal Fusion Models

Automated Red Teaming with GOAT: the Generative Offensive Agent Tester

Uncertainty-Based Abstention in LLMs Improves Safety and Reduces Hallucinations