Hacker News new | ask | show | jobs
by Ephil012 1161 days ago
How do you prevent cases where the AI might say something harmful (e.g. suggesting the user kill themselves)?