Both Anthropic and OpenAI mention the possibilities of slowing AI development in their latest "what comes next" in AI posts， but say they need to be an action coordinated across the entire world using as-yet-unidentified methods.