A Practical Memory Injection Attack against LLM Agents: 2025

Research question

Can attackers implant malicious records in an agent's long-term memory through ordinary interaction?

Key findings

MINJA injects bridging steps via queries so that later users trigger malicious reasoning - High injection and attack success with only query access

Sources

arxiv.orghttps://arxiv.org/abs/2503.03704

Contribute

Spot an error, or know a record we should add? Members can propose additions and submit corrections.

Submit research Report a correction or request removal

Related in Research Paper Library

Evaluating Coding Agents on Kernel Exploit Generation

Coding agents now find real vulnerabilities in production software. However, bug discovery results do not measure whether agents can construct exploit primitives. We introduce KEX-bench, a benchmark for evaluating…

Agent security and tool use2026

A2M: Trace-Optimized Agent Hijacking in the MCP Ecosystem

Agents using the Model Context Protocol (MCP) rely on semantic matching to select tools from third-party servers, exposing a semantic supply-chain risk through attacker-controlled metadata and outputs. We introduce…

Agent security and tool use2026

MobileCybench: Evaluating Agent Vulnerability Discovery via Executable Probes

AI agents now report vulnerabilities faster than maintainers can review them. Reports often depend on security properties specific to the application, and require considerable human labor to process. To mitigate…

Agent security and tool use2026