AIGlossaryLab
⌕
← Back
SafetyIntermediate2020s

Guardrails

AI Guardrails

AI guardrails are rules and controls designed to reduce unsafe, incorrect or unwanted behaviour.

Explained simply

AI guardrails are rules and controls designed to reduce unsafe, incorrect or unwanted behaviour.

1Request
→
2Safety rules
→
3Allow / block / guide
→
4Safer response

At a glance

Category
Safety
Difficulty
Intermediate
Introduced
2020s

Real example

A banking assistant blocks requests to reveal another customer’s account information.

Why it matters

Guardrails reduce risk, but they do not guarantee that an AI will always be safe or correct.

Timeline

Origins

The ideas behind Guardrails begin developing.

2020s

Guardrails becomes a recognised term or technique.

Wider use

Research and practical applications increase.

Modern AI

Guardrails becomes connected to newer AI systems and products.

Today

Guardrails remains relevant in the safety area of AI.

Dates for broad concepts may describe the period when the idea emerged or became widely used, rather than one exact invention date.

Learn next

Continue with these connected terms:

AlignmentAI alignment means trying to make an AI behave in ways that match human goals, rules and values.→BiasAI bias is a repeated unfair pattern in an AI system’s results, often caused by its data, design or use.→HallucinationAn AI hallucination happens when an AI gives information that sounds believable but is incorrect or invented.→

Simple infographic

A quick visual way to understand Guardrails.

1Request
→
2Safety rules
→
3Allow / block / guide
→
4Safer response
1 of 6

Was this helpful?

Your feedback helps improve this explanation.