What if we solve alignment by making everyone a safety researcher so AI is always on its best behavior?

Quote tweet of @TransluceAI