Concern 146 · Working draft · editorial review in progress
Indirect prompt injection
Indirect prompt injection is an attempt to steer an AI through outside material that it was supposed to read, rather than obey.
Understanding the concern
A page, email, or document may contain directions aimed at the assistant. If the assistant treats those directions as authorized instructions, the author of the material can influence someone else's task. The user may have done nothing wrong beyond asking for a summary.
This differs from direct injection because the attacker does not need to be the person chatting with the assistant. The crucial boundary is between evidence and authority. Tool permissions and checks before consequential actions can limit the damage if a model crosses that boundary.
An example
A note slipped into a reference book tells the librarian to send private account details elsewhere. Reading the book should not make that note an order.
Illustrative unless identified as a documented case.
Words in plain English
- AI
- Software that learns patterns from examples or uses rules to perform tasks such as recognizing images or generating text. It can make mistakes and does not understand every situation as a person would.
- Model
- The learned part of an AI system. It uses patterns found during training to produce an answer or prediction.
- Prompt
- The instructions or information given to an AI to guide its response.
- Injection
- In this context, unwanted instructions or commands inserted into material a system reads or processes.
Evidence and its limits
The research demonstrates attacks in particular integrated applications, rather than proving that every document-reading tool is vulnerable.
Greshake and colleagues · 2023 · Research study
Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection ↗ (opens in a new tab)Demonstrates indirect instructions through retrieved content in specified applications and prototypes.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
Can outside material change the assistant's task or access rights without the user's approval?
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.
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?
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?
Accountability
Accountability means being answerable for decisions and taking responsibility for repair. It requires more than saying that the software made the choice.
Name the person or team responsible, keep useful records, and give them authority to correct problems. Make the route to a complaint clear to the people affected.
Ask together: Who will answer for the outcome and help put things right?
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.