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.
Shared entity: Poisoned / Same source domain / Shared topic / What happened next
Both cover Poisoned; reported by the same outlet (arxiv.org); overlapping topics (agent, attack, during).
Shared entity: Unlike / Same source domain / Shared topic / What happened next
Both cover Unlike; reported by the same outlet (arxiv.org); overlapping topics (agent, explicit, memory).
Shared entity: Poisoned / Same source domain / Shared topic / Earlier coverage
Both cover Poisoned; reported by the same outlet (arxiv.org); overlapping topics (agent, attack, tool).
Shared entity: Unlike / Same source domain / Shared topic / What happened next
Both cover Unlike; reported by the same outlet (arxiv.org); overlapping topics (agent, memory).
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (against, agent, control, tool); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (against, agent, execution, memory); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (against, agent, attack, tool); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (against, agent, attack, tool); pushes against this story (against).