pretty strong disagree; Apple can't afford to potentially start an AI apocalypse because it tried to launch an OpenClaw type service without making it impossible for prompt-injection or agent identity hijacking to happen as we're seeing with Moltbook

Let OpenClaw experiment and beta test with the hackers who won't mind if things go sideways (risk of creating Skynet aside), and once we've collectively figured out how to create such a system that can act powerfully on behalf of its users but with solid guardrails, then Apple can implement it.