Concern 273 · Working draft · editorial review in progress
Persistent-memory poisoning
Persistent-memory poisoning changes an agent's saved knowledge or past examples so that future tasks are influenced by untrustworthy material.
Understanding the concern
An assistant may retrieve stored notes when deciding what to do. If those notes contain false facts, misleading examples, or malicious directions, later decisions can be affected even after the original interaction ends. The model's weights need not change.
This overlaps directly with persistent memory injection. The distinction is that poisoning includes corrupted facts and demonstrations, not only explicit instructions. These are overlapping ways stored information can mislead an assistant. Stored information needs origin records, access controls, and a way to correct or remove it.
An example
A shared notebook is quietly altered, and later workers rely on the changed entry as if it were a verified record.
Illustrative unless identified as a documented case.
Words in plain English
- Agent
- An AI system that can choose and carry out several steps toward a goal, using the tools and access people give it.
- Model
- The learned part of an AI system. It uses patterns found during training to produce an answer or prediction.
- Data poisoning
- Deliberately changing training material to push a system toward unwanted behavior.
- Model weights
- Numbers adjusted during training that help determine how a model responds. They are part of what makes a trained model work.
- Injection
- In this context, unwanted instructions or commands inserted into material a system reads or processes.
Evidence and its limits
AgentPoison demonstrates memory and knowledge-base poisoning in tested systems; the distinction from instruction-only injection is conceptual, not a separate prevalence estimate.
Chen and colleagues · 2024 · Research study
AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases ↗ (opens in a new tab)Demonstrates poisoned retrieval and memory in evaluated agents without requiring model retraining.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
Which saved memories are trusted, who can change them, and how can their influence be removed?
Keep track of different views, what needs checking, and any next step in your own discussion notes.
Open discussion notes →Principles for the conversation
Before you finish · reflect together
Virtues and principles to consider
Virtues are qualities we practice in how we treat people and make choices. They are not scores for judging someone. Use these ideas to explore what a thoughtful response could look like.
Prudence
Prudence means using careful judgment before and during action. It balances worthwhile opportunities with the need to avoid harm that could have been anticipated.
Start with a limited trial, set stopping points, and prefer choices that can be undone when uncertainty is high. Review what happens rather than assuming the original plan remains sound.
Ask together: What is a careful next step, and when should we stop or change course?
Humility
Humility means recognizing the limits of our knowledge and being willing to learn from others. Expertise can be valuable without answering every question.
State where evidence is weak, invite criticism, and listen to people with different experience. Leave room to revise a decision when new information appears.
Ask together: What might we be missing, and who could help us see it?
Stewardship
Stewardship means caring for resources and systems that others depend on, including people who will use them later. It looks beyond the benefit of a quick launch.
Plan for maintenance, energy use, data protection, and eventual shutdown. Ask who will look after the tool when the original team moves on and how future users can avoid inheriting hidden costs.
Ask together: What are we responsible for caring for over time?
Editorial ratings
These are editorial judgments from the register, not measured probabilities or predictions about your situation.
About the register and scoring →More ways to explore
Search for more writing
For more specific or recent material, open a search below or ask for a web summary. AI-generated summaries can contain mistakes; open the original articles to check them.