Section 112 · Chapter 14, Frontier Safety and Containment

Testing Containment Systems

Every useful AI containment system has a paradox at the center: if the system is valuable, someone or something has to interact with it. Every interaction is a possible channel.

containment

What to do

  1. Start with a channel inventory.
  2. Ask whether the AI can persuade an operator to paste text into a tool.
  3. Ask whether generated code includes hidden network calls or dependency changes.
  4. Ask whether logs reveal secrets.
  5. Ask whether a timing pattern can transmit bits.

Evidence to preserve

  • Preserve the inputs, versions, configurations, raw outcomes, and results for containment needed to reproduce work on Testing Containment Systems.
  • Report results for containment by relevant slice, separate blocker failures from averages, state uncertainty and blind spots, and connect the result to a release decision.

Expert note

In a real release review, containment testing is a systems-security problem plus a human-factors problem plus an incentive problem. Treat the AI as a strategic actor if the risk model requires it.

Continue the conversation

Apply this to your context.

Save your product context once, then open a focused conversation that combines it with this concept.

Cite this page

Jason Arbon. "Testing Containment Systems." Testing AI Knowledge Edition, section 112.

https://jarbon.ai/testing-ai/knowledge/ch112-containment.html

Shared across the Knowledge Edition

Adapt every concept to your world.

Describe your product, role, users, risks, constraints, or current quality problem. This stays in this browser until you choose to send it to ChatGPT.

Saved only in this browser.0 / 2400