Threats and defensive boundaries specific to LLM systems, including untrusted instructions, compromised evidence, and constrained actions.
Tagged articles
11 articles in this tag.
Two permission grants land on the review table. One lets the agent read every ticket and send email. The other lets it read assigned tickets and draft…
Read tutorial
The moment you give an agent a tool, you stop deciding what it does and start deciding what it is allowed to do. The real design question is not "can the…
Read tutorial
If you've spent any time reading about AI security, you've probably seen "jailbreak" and "prompt injection" used interchangeably. The terms blur because…
Read tutorial
That single distinction separates a demo from a system. When you chat with an LLM directly, you are the pipeline: you judge the output, check the facts,…
Read tutorial
When people first hear that an LLM can "use tools," they usually picture the model reaching out and doing something itself—searching the web, running code,…
Read tutorial
You can recite that untrusted content can carry instructions. Then someone shows you a proposed action and asks, "Should this run?" and the recitation…
Read tutorial
A proposed tool call sits on your screen: send the summary to the client. You have a policy document open in another tab. Is this allowed, does it need…
Read tutorial
A citation chip proves a chunk was retrieved. It does not prove the chunk was ever authorized or trustworthy.
Read tutorial
Your RAG app is working beautifully. Users ask questions, your system retrieves the right documents, and the model produces clear, grounded answers. Then…
Read tutorial
A team ships an assistant that reads inbound email, searches a knowledge base, and can send replies. Then the argument starts. "We added a guardrail." "The…
Read tutorial
A support chatbot answers a customer's question with total confidence. The answer cites a document from the company knowledge base. The only problem: the…
Read tutorial