Latest work by our policy team analyzes the game theory of industry cooperation on AI safety:
I think that AI is more likely to be broadly beneficial if the AI community takes a cooperative approach to its safety, security, and social impact. Excited to hear from others interested in this work, or in developing and implementing cooperative strategies.