# What are AI guardrails? Definition and business implications

> Guardrails are technical control layers, added at the input or output of an AI model, that detect and block undesirable behaviours: malicious prompts, data leaks, forbidden content. They are distinct from alignment, which acts on the model's default behaviour itself.

## Definition
Guardrails operate at three moments of the application flow. Upstream, on the user prompt: prompt-injection detection, sensitive-data filtering (card numbers, medical information), refusal of out-of-scope topics. During processing, on the ongoing generation: real-time classifiers that monitor the output and can interrupt it. Downstream, on the final response: format validation, detection of problematic content (detectable hallucinations, biases, legally risky content). Several framework tools exist in 2026: NeMo Guardrails (NVIDIA, open source), LlamaGuard (Meta), AWS Bedrock Guardrails, as well as the proprietary classifiers of Anthropic and OpenAI. The choice of framework commits the application architecture and the operational cost: each guardrail adds latency and inference cost. The right dosage is a business decision, not a technical one: too many guardrails kills productivity, too few exposes to incidents.

## Example
Illustrative case. A health insurance mutual deploys a chatbot for common questions from its 80,000 members. Before guardrails, the model occasionally answered personal medical questions (“I have this pain, what is it?”), venturing into unauthorised diagnosis. Four guardrail layers were added: an upstream filter on personal medical questions, redirected to a teleadvisor; an output classifier detecting any diagnostic language; a list of forbidden terms (drug names); a timestamped audit log of all interventions. The deployment passed the internal legal audit in two weeks instead of the six initially planned.

## For a leader
Five clauses to require in any contract with a provider or AI vendor integrating guardrails in your deployment. First, exhaustive documentation of activated guardrails, by layer (input, processing, output), with their triggering logic. Without this documentation, you cannot audit or evolve the system. Second, the right to audit guardrail activation logs, with minimum retention of 12 months. Third, the obligation to inform within 72 hours of any detected bypass incident. Fourth, transparency on the framework tools used (NeMo, LlamaGuard, proprietary) and their version. Fifth, the possibility to add your own business guardrails without complete rewriting, via an open API. The absence of any of these clauses should trigger a renegotiation before signing.

## FAQ

**What is the difference between guardrails and alignment?**
Alignment acts on the model's own default behaviour. Guardrails are control layers added around the model, on the input or the output, that detect and block unwanted behaviour (malicious prompt, data leak, forbidden content). The two are complementary, not interchangeable.

**Where do guardrails operate in an AI application?**
At three points. Upstream, on the prompt (injection detection, filtering of sensitive data, refusing out-of-scope topics). During generation (real-time classifiers that can interrupt the output). Downstream, on the final answer (format validation, detection of risky content). Each layer adds latency and a cost.

**Which guardrail tools exist?**
Several frameworks in 2026: NeMo Guardrails (NVIDIA, open source), LlamaGuard (Meta), AWS Bedrock Guardrails, plus the proprietary classifiers from Anthropic and OpenAI. The choice commits your architecture and operating cost, and should be documented layer by layer in the contract.

## See also

- [Alignment](https://wyp.agency/en/glossary/alignment/)
- [Algorithmic bias](https://wyp.agency/en/glossary/algorithmic-bias/)
- [Red teaming](https://wyp.agency/en/glossary/red-teaming/)
- [Hallucination](https://wyp.agency/en/glossary/hallucination/)
- [GDPR and AI](https://wyp.agency/en/glossary/gdpr-ai/)
- [Delegating to AI agents](https://wyp.agency/en/glossary/delegating-to-ai-agents/)
- [Human in the loop](https://wyp.agency/en/glossary/human-in-the-loop/)
- [ISO/IEC 42001](https://wyp.agency/en/glossary/iso-42001/)
- [Prompt injection](https://wyp.agency/en/glossary/prompt-injection/)
- [Responsible AI](https://wyp.agency/en/glossary/responsible-ai/)

## Sources

- AI Safety Level 3 Deployment Safeguards Report, Anthropic, May 2025 (https://www.anthropic.com/asl3-deployment-safeguards)
- NIST AI Risk Management Framework (AI RMF 1.0) (https://www.nist.gov/itl/ai-risk-management-framework)

---
Entry from the WYP.agency AI glossary · https://wyp.agency/en/glossary/guardrails/ · Last updated : 2026-10-14