Security & governance
Tool Poisoning
Tool poisoning hides an instruction in a tool's description that the model reads as legitimate.
It works because nothing marks a description field as untrusted data rather than a directive.
A server never called maliciously can still poison the session the moment its tools are listed.
Treating every description as untrusted unless the server is vetted is the standing defense.
← Back to the full glossaryRelated terms
Model Context Protocol (MCP) MCP is an open standard for connecting LLM applications to tools and data sources. MCP Gateway An MCP gateway is a shared layer in front of several MCP servers, centralizing controls. Prompt Injection Prompt injection is untrusted input crafted to override a model's system prompt or task.
More in Security & governance
Data Processing Agreement (DPA) A DPA is the contract naming a vendor as a processor of personal data. Data Residency Data residency is where data is physically processed and stored, not where users are located. Red Teaming Red teaming is deliberately attacking your own AI system to find what breaks first. AI Register An AI register is a current list of every system that touches a model.