Safety & Alignment

Refusal

Quick Answer

When an LLM declines to respond to a request, typically for safety or ethical reasons.

Refusal is when models decline requests. Models should refuse harmful requests. Appropriate refusals: 'I can't help with that'. Refusal patterns are learned during training. Over-refusal (refusing legitimate requests) is problematic. Balancing safety and utility is challenging. Refusal calibration is important. Refusal is a key safety mechanism.

Last verified: 2026-04-08

Compare models

See how different LLMs compare on benchmarks, pricing, and speed.

Browse all models →