Known By Their Actions: Fingerprinting LLM Browser Agents via UI Traces
This work highlights a new attack surface for LLM agents, enabling targeted attacks based on model identity, which is a security concern for agent deployment.
The authors show that passive observation of LLM-based browser agents' actions and interaction timings can identify the underlying model with up to 96% F1 across 14 models and 4 web environments, posing a security risk.
As LLM-based agents increasingly browse the web on users' behalf, a natural question arises: can websites passively identify which underlying model powers an agent? Doing so would represent a significant security risk, enabling targeted attacks tailored to known model vulnerabilities. Across 14 frontier LLMs and four web environments spanning information retrieval and shopping tasks, we show that an agent's actions and interaction timings, captured via a passive JavaScript tracker, are sufficient to identify the underlying model with up to 96\% F1. We formalise this attack surface by demonstrating that classifiers trained on agent actions generalise across model sizes and families. We further show that strong classifiers can be trained from few interaction traces and that agent identity can be inferred early within an episode. Injecting randomised timing delays between actions substantially degrades classifier performance, but does not provide robust protection: a classifier retrained on delayed traces largely recovers performance. We release our harness and a labelled corpus of agent traces \href{https://github.com/KabakaWilliam/known_actions}{here}.