ExplainerSteady· Updated 6 hours ago
Prompt injection
A trick where hidden instructions in a web page, email or file make an AI follow an attacker's orders. It is the main security risk for agents that read outside content.
Why it matters
Anyone building agents is expected to know how to limit the damage a fooled model can do.
#security#agents
More on the radar
BuzzwordRisingTop story
Model Context Protocol (MCP)
MCP is an open standard that lets AI apps connect to tools and data in one common way, so nobody has to build a custom link for every app and tool pair.
BuzzwordRising
Agentic AI
Agentic AI plans and takes steps on its own, using tools, instead of only answering one question at a time. It can search, decide, act and check its work.
BuzzwordRising
Context engineering
The skill of choosing what a model sees for a task: the instructions, the right documents, the memory and the tool results, and leaving out the rest.