Lessons learned on language model safety and misuse

3 years ago 10

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