Open-world agents suffer from an Authorization-Execution Gap arising from delegation incompleteness, channel corruption, and composition fragmentation, requiring dynamic runtime integrity checks instead of only upfront filters or post-hoc audits.
When benign inputs lead to severe harms: Eliciting unsafe unintended behaviors of computer-use agents
4 Pith papers cite this work. Polarity classification is still indexing.
citation-role summary
citation-polarity summary
years
2026 4representative citing papers
Routine user chats can unintentionally poison the long-term state of personalized LLM agents, causing authorization drift, tool escalation, and unchecked autonomy, as measured by a new benchmark and reduced by the StateGuard defense.
PageGuide grounds LLM answers in webpage DOM elements using visual overlays for find, guide, and hide modes, yielding measurable gains in a 94-user study.
The paper develops a unified framework that organizes computer-use agent reliability around perception-decision-execution layers and creation-deployment-operation-maintenance stages to map security and alignment interventions.
citing papers explorer
-
The Authorization-Execution Gap Is a Major Safety and Security Problem in Open-World Agents
Open-world agents suffer from an Authorization-Execution Gap arising from delegation incompleteness, channel corruption, and composition fragmentation, requiring dynamic runtime integrity checks instead of only upfront filters or post-hoc audits.
-
When Routine Chats Turn Toxic: Unintended Long-Term State Poisoning in Personalized Agents
Routine user chats can unintentionally poison the long-term state of personalized LLM agents, causing authorization drift, tool escalation, and unchecked autonomy, as measured by a new benchmark and reduced by the StateGuard defense.
-
PageGuide: Browser extension to assist users in navigating a webpage and locating information
PageGuide grounds LLM answers in webpage DOM elements using visual overlays for find, guide, and hide modes, yielding measurable gains in a 94-user study.
-
Securing Computer-Use Agents: A Unified Architecture-Lifecycle Framework for Deployment-Grounded Reliability
The paper develops a unified framework that organizes computer-use agent reliability around perception-decision-execution layers and creation-deployment-operation-maintenance stages to map security and alignment interventions.