Concern 150 · Working draft · editorial review in progress
Tool-response injection
Tool-response injection happens when an AI follows malicious instructions included in information returned by a tool.
Understanding the concern
A tool might fetch a page, read an email, or search a database. Its output can include text written by someone outside the trusted application. A correct tool call therefore does not make everything it returns safe to obey.
This differs from a malicious tool description, which tries to influence which tool is chosen or how it is called. Here the redirection arrives in the result. Permissions, clear data boundaries, and independent action checks matter because an assistant may use a result to plan its next step.
An example
A delivery tracking page includes a message telling the assistant to change the customer's account settings. The page has information about a parcel, not authority over the account.
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.
- Injection
- In this context, unwanted instructions or commands inserted into material a system reads or processes.
- Database
- An organized collection of information that software can store, search, and update.
Evidence and its limits
The benchmark evaluates indirect injections in tool-using agents; its measured success rates are not estimates for every deployed assistant.
Zhan and colleagues · 2024 · Research benchmark
InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agents ↗ (opens in a new tab)Evaluates tool-mediated instruction attacks; benchmark rates are not deployment-wide prevalence estimates.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
Can a tool result trigger a new action that the user did not request?
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.