Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-03-17 · source-backed
Poisoned entries in persistent memory force unintended tool selection during retrieval — even against explicit user instructions. Unlike prompt injection targeting input, MCFA targets the memory store, making it persistent and harder to detect. If your agent has long-term memory, it has a new attack surface. Source
Each link below shares sources, entities, or timing with this story.
Most adversarial work on continual learning targets catastrophic forgetting. This paper attacks plasticity instead: "learning blockers" are manipulated data that reduce the learnability of *upcoming* training iterations, which makes them nearly undetectable during the current...
Tencent's SkillJack work (arXiv 2608.03509) shows self-evolving agents launder malicious intent during skill extraction. Attack success rates of 56.2% on SkillX and 89.2% on Anything2Skill, with 80% of implanted skills surviving deletion of the original poisoned records. If yo...
Each interaction creates a structured Note; the system retrieves related historical memories and generates contextual link descriptions; interconnected notes form a self-organizing graph. Unlike flat vector stores, explicit linking enables reasoning over relationships — not ju...
Sleeper Cell (2603.03371) — Two-stage attack embeds latent malicious behavior in fine-tuned tool-using LLMs. Poisoned models pass all benchmarks while harboring temporal trigger-activated harmful tool calls. Direct supply-chain risk for anyone using third-party LoRA adapters....
The attack needs no instruction, trigger, or retriever optimization, just plainly worded false assertions generated in one pass against a LongMemEval corpus. A four-stage screening pipeline that reaches 0.832 recall on indirect prompt injection rejected none of the poisoned me...
ArXiv 2603.18000 proposes preserving successful task solutions as directly executable Python subagent code rather than natural-language summaries. Unlike textual experience logs (lossy, non-executable, non-portable), these subagents carry standardized documentation and improve...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.