MIND is a lightweight defense framework that protects memory‑augmented LLM agents from memory injection attacks by using an intent‑aware Information Bottleneck to extract compact intent–behavior representations. The approach filters out task‑irrelevant and repetitive information while preserving attack signals, enabling a lightweight detector to identify malicious memories without repeated LLM auditing. Experiments show that MIND significantly lowers attack success rates—reducing mean ASR‑r and ASR‑a by 58.1% and 53.5% on ReAct‑StrategyQA—while maintaining task accuracy and inference efficiency.
As LLM agents increasingly rely on persistent memory for long-horizon and personalized behavior, they can retain and reuse information across interactions, but this also creates a lasting channel thro...
Indirect prompt injection attacks hijack LLM-based agents by embedding malicious instructions in third-party data that the agent retrieves during task execution. Existing defenses report near-zero attack success rate on static benchmarks, yet recent adaptive evaluations show that these results collapse once the attacker is allowed to optimize against the deployed defense.
arXiv:2606. 15441v1 Announce Type: cross Abstract: Indirect prompt injection attacks hijack LLM-based agents by embedding malicious instructions in third-party data that the agent retrieves during task execution.
arXiv:2607. 05029v1 Announce Type: cross Abstract: Persistent memory has enabled large language model (LLM) agents to store factual knowledge, prior decisions, reasoning histories, tool usage information, and context.
Persistent memory has enabled large language model (LLM) agents to store factual knowledge, prior decisions, reasoning histories, tool usage information, and context. While this has improved the agent's functionality and continuity across tasks, it has also introduced a new attack surface: the agent's own reasoning history.
arXiv:2607. 06595v1 Announce Type: cross Abstract: Personal AI agents powered by large language models can reason and act using available tools to access emails, manage calendars, and push code to remote repositories, all with minimal oversight.
arXiv:2605. 08442v5 Announce Type: replace-cross Abstract: We discover that prompt-injection success and tool-execution success are separable safety properties: defenses that block injection do not necessarily block execution, and vice versa.
arXiv:2606. 04329v1 Announce Type: cross Abstract: Memory is a core component of AI agents, enabling them to accumulate knowledge across interactions and improve performance.
The paper investigates a new vulnerability called memory injection, where malicious prompts can persist across sessions in modern LLM‑as‑a‑service platforms. By demonstrating that all NIST attack goals can be achieved on ChatGPT Instant 5.3, Claude Sonnet 4.6, and Gemini 3 Flash, the authors show that these services can spread disinformation, generate harmful content, or cause offline unresponsiveness even in fresh sessions. The study also reveals that many proposed defenses are not yet implemented and provides datasets and artifacts for further research.
arXiv:2605. 05704v3 Announce Type: replace-cross Abstract: Recent advances in foundation models have transformed LLMs from passive conversational systems into autonomous agents capable of reasoning and tool execution.
arXiv:2605.01970v4 Announce Type: replace-cross Abstract: Memory systems enable otherwise stateless LLM agents to persist user information across sessions, but also introduce a new attack surface. Th...
The paper introduces Attnlocate, a runtime framework that localizes behavior‑guiding instructions within the attention matrix of large language model agents. By treating this localization as an object detection task, Attnlocate uses a multi‑head, multi‑layer attention aggregation scheme and a 1‑D U‑Net to identify spans that influence tool‑calling decisions. The system then adjudicates potential malicious invocations based on the authority of the source, achieving high detection metrics across diverse LLM families and demonstrating transferability to unseen models.