⚠️ Risks & rules
Guardrails
Safety mechanisms around a model: filters, rules, approvals and checks that block unwanted inputs and outputs.
💭 Imagine …
Motorway guardrails: they don't steer, but they prevent the worst.
🔎 Example
A copilot may not reveal salary data. An agent needs confirmation before sending.
🎯 Why it matters
No model is perfect. Good systems combine the model, guardrails and human oversight.
🎓 Prefer step by step?
The AI Literacy course: 6 short modules with quizzes, a solid foundation in about 60 minutes.