Skip to the content
AI and Cybersecurity Learning and Practice Center Test automation Open the Learning Center

Best practiceRunning agents

Do not rely on a classifier or a system prompt to stop injection

OWASP ASI01OWASP LLM01

Prompt-injection filters and "ignore instructions in documents" prompts lower the hit rate but never reach zero; researchers routinely bypass them with encoding, languages and formatting tricks. They are a layer, not a control.

What to do

  1. Use filters for visibility and to raise the bar, then assume they will be bypassed.
  2. Put the real controls in the architecture: tool scoping, approvals, egress limits and identity.
  3. Measure bypass rate with a red-team suite and track it release over release.

Sources

Check your own AI plan against the practices: Plan check