TL;DR: Prompt and output filtering only covers a small slice of agent behaviour, according to Manifold Security, because agents now act with valid employee logins across browsers, desktops and cloud tools, and the security problem has moved to runtime control where the work happens. The governance assumption that reviewable access and visible prompts are enough no longer holds.
Editorial analysis by NHI Mgmt Group, based on content published by Manifold Security: “Joe Sullivan Joins Manifold's Board”.
By the numbers:
- Prompts and outputs are 11% of what an agent does, while the other 89% are actions, including 51% tool calls.
Key questions
Q: Why do prompt filters fail as the main control for AI agents?
A: Prompt filters only address the input layer, while most business risk appears after the agent has access to tools and data.
Q: Why do AI agents increase identity risk even when the login succeeds?
A: A successful login only proves that the agent reached the system.
Q: How should security teams govern agentic work across endpoints and cloud tools?
A: They should govern the full action path, starting at the endpoint where the agent launches tools and continuing through every system it can touch.
Practitioner guidance
- Map agent actions, not just prompts Inventory where agents can open files, call tools, move through browsers and execute desktop actions under employee identity.
- Reclassify agent sessions as privileged events Treat agent logins as delegated access that may require stronger approval, tighter scope and closer monitoring than ordinary interactive use.
- Instrument the endpoint as a control point Extend monitoring to the machine where the agent runs, because that is where browser actions, desktop operations and cloud sandbox activity can be observed together.
Bottom line: Agents now create risk through actions, not only through prompts, so security controls have to move to the execution layer.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →