TL;DR
Genesis introduces an evolving, agentic framework for attacking web-based LLM agents, leveraging genetic algorithms and dynamic strategy learning to improve attack success over static methods.
Contribution
The paper presents a novel framework combining genetic algorithms and dynamic strategy learning for continuous evolution of attack strategies against web LLM agents.
Findings
Outperforms existing attack baselines in various web tasks.
Discovers novel attack strategies through continuous evolution.
Effectively adapts to diverse web agent behaviors.
Abstract
As large language model (LLM) agents increasingly automate complex web tasks, they boost productivity while simultaneously introducing new security risks. However, relevant studies on web agent attacks remain limited. Existing red-teaming approaches mainly rely on manually crafted attack strategies or static models trained offline. Such methods fail to capture the underlying behavioral patterns of web agents, making it difficult to generalize across diverse environments. In web agent attacks, success requires the continuous discovery and evolution of attack strategies. To this end, we propose Genesis, a novel agentic framework composed of three modules: Attacker, Scorer, and Strategist. The Attacker generates adversarial injections by integrating the genetic algorithm with a hybrid strategy representation. The Scorer evaluates the target web agent's responses to provide feedback. The…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
