When Consent Outlives Context: Residual Authority Replay in Long-Lived Agents
cs.AI
Submitted: 2026-09-27
Updated: 2026-09-27
Terminology
Sources
- TAME: A Trustworthy Test-Time Evolution of Agent Memory with Systematic Benchmarking
- When Context Gets Root: Privilege Escalation in LLM Harnesses
- Options, Not Clicks: Lattice Refinement for Consent-Driven MCP Authorization
- Practice Makes Unsafe: Skill Misevolution in Self-Improving LLM Agents
- MemGPT: Towards LLMs as Operating Systems
- FragFuse: Bypassing Access Control of Large Language Model Agents via Memory-Based Query Fragmentation and Fusion
- Lingering Authority: Revocable Resource-and-Effect Capabilities for Coding Agents
- Temporary Authority, Permanent Effects: Commit-Time Authorization for LLM Agents
- Beyond OAuth: Task-Scoped Authorization for AI Agents via Natural Language Slices
- Voyager: An Open-Ended Embodied Agent with Large Language Models
- Safe to Resume? Breaking Execution Continuity of Agent Execution via Rollback
- Qwen3 Technical Report
- When Lower Privileges Suffice: Investigating Over-Privileged Tool Selection in LLM Agents
- Zombie Agents: Persistent Control of Self-Evolving LLM Agents via Self-Reinforcing Injections
- Are You Still the Agent I Authorized? Earned Authority under a Fixed Ceiling for Evolving Agents
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection