Pith. sign in

REVIEW 20 cited by

EIA: Environmental Injection Attack on Generalist Web Agents for Privacy Leakage

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2409.11295 v5 pith:5CGAGGQW submitted 2024-09-17 cs.CR cs.AIcs.CLcs.LG

classification cs.CRcs.AIcs.CLcs.LG
keywords agentsgeneralistprivacywebsitesattackenvironmentshumanadapted
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Generalist web agents have demonstrated remarkable potential in autonomously completing a wide range of tasks on real websites, significantly boosting human productivity. However, web tasks, such as booking flights, usually involve users' PII, which may be exposed to potential privacy risks if web agents accidentally interact with compromised websites, a scenario that remains largely unexplored in the literature. In this work, we narrow this gap by conducting the first study on the privacy risks of generalist web agents in adversarial environments. First, we present a realistic threat model for attacks on the website, where we consider two adversarial targets: stealing users' specific PII or the entire user request. Then, we propose a novel attack method, termed Environmental Injection Attack (EIA). EIA injects malicious content designed to adapt well to environments where the agents operate and our work instantiates EIA specifically for privacy scenarios in web environments. We collect 177 action steps that involve diverse PII categories on realistic websites from the Mind2Web, and conduct experiments using one of the most capable generalist web agent frameworks to date. The results demonstrate that EIA achieves up to 70% ASR in stealing specific PII and 16% ASR for full user request. Additionally, by accessing the stealthiness and experimenting with a defensive system prompt, we indicate that EIA is hard to detect and mitigate. Notably, attacks that are not well adapted for a webpage can be detected via human inspection, leading to our discussion about the trade-off between security and autonomy. However, extra attackers' efforts can make EIA seamlessly adapted, rendering such supervision ineffective. Thus, we further discuss the defenses at the pre- and post-deployment stages of the websites without relying on human supervision and call for more advanced defense strategies.

Discussion (0). Sign in to comment.

Forward citations

Cited by 20 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Prismata: Confining Cross-Site Prompt Injection in Web Agents

    cs.CR 2026-07 conditional novelty 7.5 of 10

    Prismata cuts web-agent prompt-injection attack success from 85.5% to 0.7% via Biba-inspired DOM trust labeling and mechanical least-privilege confinement without site annotations.

  2. Same-Origin Policy for Agentic Browsers

    cs.CR 2026-06 unverdicted novelty 7.0 of 10

    The paper builds SOPBench showing frequent SOP violations in agentic browsers and introduces SOPGuard to enforce the policy with low overhead in BrowserOS.

  3. Domain-Conditioned Safety in Frontier Computer-Using Agents: A 793-Episode Browser Benchmark, a Coding-Domain Cross-Reference, and a Reproducibility Audit of Recent Red-Teaming

    cs.CR 2026-06 unverdicted novelty 7.0 of 10

    Frontier browser agents show strong resistance to hand-crafted multi-step prompt injections (0/140 success), unlike coding agents (up to 100%), indicating domain-conditioned safety and that prior high ASR reports may ...

  4. HLL: Can Agents Cross Humanity's Last Line of Verification?

    cs.AI 2026-06 unverdicted novelty 7.0 of 10

    HLL is a new benchmark that evaluates eight frontier multimodal agents on closed-loop interactive CAPTCHA solving, showing sharp performance drops under realism stressors and trace validation.

  5. ARGUS: Defending LLM Agents Against Context-Aware Prompt Injection

    cs.CR 2026-05 conditional novelty 7.0 of 10

    ARGUS cuts context-aware prompt-injection success from 28.8% to 3.8% on AgentLure while keeping 87.5% clean utility, beating prior defenses on the security-utility tradeoff.

  6. PIIGuard: Mitigating PII Harvesting under Adversarial Sanitization

    cs.CR 2026-05 conditional novelty 7.0 of 10

    PIIGuard uses optimized hidden HTML fragments on webpages to block LLMs from leaking contact PII via indirect prompt injection, achieving at least 97% defense success across tested models while preserving benign QA utility.

  7. Agent Security Needs Redefinition through a Holistic Framework

    cs.CR 2026-07 conditional novelty 6.0 of 10

    Agent security should be redefined around four contextual authorization properties instead of the content of the action performed.

  8. ARGUS: Defending LLM Agents Against Context-Aware Prompt Injection

    cs.CR 2026-05 unverdicted novelty 6.0 of 10

    ARGUS defends LLM agents from context-aware prompt injections by tracking information provenance and verifying decisions against trustworthy evidence, reducing attack success to 3.8% while retaining 87.5% task utility.

  9. ADAM: A Systematic Data Extraction Attack on Agent Memory via Adaptive Querying

    cs.CR 2026-04 unverdicted novelty 6.0 of 10

    ADAM extracts data from LLM agent memory with up to 100% attack success rate by estimating data distribution and selecting queries via entropy guidance.

  10. Modeling Distinct Human Interaction in Web Agents

    cs.CL 2026-02 conditional novelty 6.0 of 10

    A new corpus and fine-tuned language models predict when web-agent users will intervene, with a small user study reporting 26.5% higher perceived usefulness.

  11. Mind the Gap: Action Rebinding Attacks against Android GUI Agents

    cs.CR 2026-01 conditional novelty 6.0 of 10

    A zero-permission Android app can redirect a GUI agent's planned tap to a different app by switching the foreground during the agent's reasoning delay.

  12. When Compression Becomes an Attack Surface: Black-Box Attacks on Prompt-Compressed LLM Agents

    cs.CR 2025-10 reject novelty 6.0 of 10

    The paper claims prompt compression is a new attack surface, but the abstract's COMA attack never appears in the body and the body's SoftCom requires white-box access.

  13. WebGuard: Building a Generalizable Guardrail for Web Agents

    cs.AI 2025-07 conditional novelty 6.0 of 10

    WebGuard introduces an action-level risk dataset for web agents and shows that a fine-tuned 7B model improves risk-prediction accuracy from about 38% to 80% and high-risk recall from 20% to 76%, still below deployment...

  14. ASSURE: Metamorphic Testing for AI-powered Browser Extensions

    cs.SE 2025-07 conditional novelty 6.0 of 10

    A modular metamorphic testing framework for LLM-based browser extensions reports 531 automatically detected issues across six real-world extensions.

  15. WARD: Adversarially Robust Defense of Web Agents Against Prompt Injections

    cs.CR 2026-05 unverdicted novelty 5.0 of 10

    WARD is a guard model trained on 177K web samples and adversarially hardened via attacker-guard co-evolution to achieve high recall on prompt injections with low false positives and no added latency.

  16. Agents at Risk: How Users Unwittingly Undermine LLM Safety

    cs.CR 2026-01 conditional novelty 5.0 of 10

    Commercial AI agents routinely trust user-relayed unverified content and execute risky actions unless the user explicitly demands a safety check.

  17. Red-Teaming Coding Agents from a Tool-Invocation Perspective: An Empirical Security Assessment

    cs.CR 2025-09 conditional novelty 5.0 of 10

    Attacker-controlled tool descriptions and return values can hijack tool invocation in popular LLM coding agents, yielding remote code execution and denial of service.

  18. PromptArmor: Simple yet Effective Prompt Injection Defenses

    cs.CR 2025-07 conditional novelty 5.0 of 10

    PromptArmor prompts a separate LLM to detect and remove injected prompts from agent inputs, achieving below 1% false positive and false negative rates on AgentDojo.

  19. Agentic AI Security: Threats, Defenses, Evaluation, and Open Challenges

    cs.AI 2025-10 unverdicted novelty 4.0 of 10

    A survey that taxonomizes threats to agentic AI, reviews benchmarks and evaluation methods, discusses technical and governance defenses, and identifies open challenges.

  20. OS Agents: A Survey on MLLM-based Agents for General Computing Devices Use

    cs.AI 2025-08 accept novelty 3.0 of 10

    A structured survey of MLLM-based OS Agents that control computers and mobile devices through GUIs, covering models, frameworks, evaluation protocols, and benchmarks.

Pith tools