Lessons learned on language model safety and misuse

4 years ago 12

We describe our latest thinking in the hope of helping other AI developers address safety and misuse of deployed models.