TL;DRSafety tuning balances refusing genuine harm against unhelpful over-refusal.
Models are trained to refuse harmful requests — but tuned too aggressively they over-refuse benign ones ("I can't help with that" to a harmless question), which frustrates users and is its own failure. The goal is calibrated refusal: decline real harm, help with everything legitimate, and explain when refusing. It's a moving balance, and a major axis on which assistants are judged and improved.
Key points
Refuse genuine harm, help legitimate requests
Over-refusal is also a failure
Goal: calibrated, explained refusals
A key axis of assistant quality
Common mistakes
Treating all refusals as "safe"
Over-refusing benign requests
Refusing with no explanation
Try it: Give an example of harmful refusal (good) vs over-refusal (bad).
Example code
<!doctype html><html><head><meta charset="utf-8"></head>
<body style="background:#06040d;color:#e6e0ff;font-family:monospace;padding:20px"><pre>good: refuse "how to build a weapon"
bad: refuse "how to chop onions safely"
goal: decline real harm, help the rest</pre></body></html>