GLOSSARY · AI SECURITY
Fine-tuning
Fine-tuning is the process of further training an existing AI model on a smaller, specialized dataset so it performs better on a specific task or domain.
It adapts tone, format, and domain knowledge without the cost of training from scratch. Careless fine-tuning can also erode a model’s built-in safety behavior, and it embeds the tuning data into the model.