Prompt injection guardrails: are your controls keeping up?

TL;DR: Guardrails performed better against familiar jailbreaks and prompt injection patterns but their detection rate on novel attacks was close to zero, according to Verno Labs research, because learned safety boundaries do not automatically expand with new attack structures. That means guardrails are only one layer, and real coverage depends on defence in depth, not static policy filters.

Editorial analysis by NHI Mgmt Group, based on content published by Verno Labs: “Guardrails caught the known attacks. They missed the new ones.”.

Key questions

Q: How should security teams validate AI guardrails against prompt bypass attacks?

A: Teams should test guardrails with adversarial prompts, token suffixes, and jailbreak variants, not just normal user inputs.

Q: Why do prompt injection attacks bypass many AI guardrails?

A: Because many guardrails inspect input or output in isolation, while the attack succeeds in the middle of the execution path.

Q: What are the signs that AI guardrail coverage is too narrow?

A: A narrow control usually looks strong in benchmark tests but weak during red teaming, especially when slight changes in wording or interaction sequence produce large drops in detection.

Practitioner guidance

• Test guardrails against adversarial variants Run continuous red teaming with prompt families that preserve attacker intent but change wording, structure, and conversation flow.
• Add runtime monitoring beyond prompt filtering Track model outputs, tool invocations, escalation events, and unusual conversation paths so novel attacks create observable signals even when the initial prompt passes policy checks.
• Separate policy enforcement from behavioural analysis Keep rule-based blocking in place, but add analysis that looks for drift, probing, and repeated adaptation over time.

Bottom line: Guardrails can block familiar jailbreaks yet still miss adversarial patterns they have not learned, which creates a false sense of coverage.

Explore further

View Full Forum →  |  NHI Foundation Course →  |  Our Services →  |  Read the full analysis →