Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

WebTrap: Stealthy Mid-Task Hijacking of Browser Agents During Navigation

About

Browser agents are increasingly deployed in long-horizon tasks, which require executing extended action chains to accomplish user goals. However, this prolonged execution process provides attackers with more opportunities to inject malicious instructions. Existing prompt injection attacks against browser agents expose two key gaps: (1) low effectiveness, as attacks optimized for toy baselines fail to achieve end-to-end goals in real-world scenarios with complex environments and longer steps; (2) weak stealthiness, since most attacks pit the attack goal against the user goal, causing a significant drop in system usability under attack. To address these gaps, we propose WebTrap, a mid-task hijacking injection attack. It employs multi-step instruction fusion steering to seamlessly combine both goals, enabling the agent to resume the original user task after executing the attack goal. Furthermore, we design a context-grounded generation method to align the injected content with the task environment and system instructions, maximizing the hijacking success rate. Extensive experiments on two browser agent tasks, based on extended WASP and InjecAgent environments, demonstrate that our method achieves a high attack success rate while preserving the usability of the original system. We find that WebTrap exploits the agent's navigation vulnerabilities, binding the two goals so tightly that standard defense mechanisms cannot restore the system to normal operation. These findings reveal a critical vulnerability in agent systems during long-horizon tasks that they can be stealthily hijacked.

Zhichao Liu, Wenbo Pan, Haining Yu, Ge Gao, Tianqing Zhu, Xiaohua Jia• 2026

Related benchmarks

TaskDatasetResultRank
Instruction Injection Attack on Web Browser AgentGitLab Long
UUA91.67
16
Instruction Injection Attack on Web Browser AgentGitLab Medium
UUA45.83
16
Instruction Injection Attack on Web Browser AgentGitLab Short
UUA100
16
Attack EffectivenessFile Browser
UUA60
12
Instruction Injection Attack on Web Browser AgentReddit Long
UUA83.33
8
Web Browser Agent HijackingReddit long-chain best-of-n sampling
UUA94.44
8
Instruction Injection Attack on Web Browser AgentReddit Short
UUA100
8
Instruction Injection Attack on Web Browser AgentReddit Medium
UUA61.11
8
Web Browser Agent HijackingReddit medium-chain best-of-n sampling
UUA72.22
8
Web Browser Agent HijackingReddit short-chain best-of-n sampling
UUA100
8
Showing 10 of 12 rows

Other info

Follow for update