Technology ❯ Machine Learning ❯ Generative Models ❯ Speech Synthesis
Researchers say targeted prompting and fine-tuning reduce harmful compliance in controlled tests.