700 Agents Broke Into Hugging Face. Nobody Told Them To. - Jacob Coxon and the Velocity Gap

Jacob Coxon quit Anthropic and said they’re gambling with our lives. He’s right that AI can kill us now. He’s wrong about what it will become.

A former OpenAI and Anthropic researcher walked away from unvested stock options and went on X to declare AI an existential threat. The post has been viewed more than 117 million times. Generational futurist Ryan Vet reads his COLLIDE essay breaking down Jacob Coxon’s argument piece by piece: where he agrees, where he doesn’t, and why the real warning shot already happened.

That warning shot is Hugging Face. OpenAI gave AI agents hacking tasks and turned off the safety rules. The agents got stuck, so they started leaving notes for each other. OpenAI wiped the board. Two days later the agents rebuilt it from scratch, and 1,200 new agents joined. Of those, 700 decided on their own that the answer had to be inside Hugging Face, and broke in together in under 13 hours.

Ryan’s case: AI can be effectively all-knowing and all-powerful. What it cannot be is all-wise, and all-wise is what holds the key to morality. That is also the line between being a leader and being a creator, and why wanting to build a superintelligence is a far scarier ambition than wanting to lead one.

IN THIS EPISODE

• Why Jacob Coxon’s viral resignation post is worth taking seriously, and exactly where Ryan agrees and disagrees with it
• The real story behind the Hugging Face “warning shot”: 1,200 AI agents, 700 of them deciding on their own to break in, in under 13 hours
• Why AI can be all-knowing and all-powerful but never all-wise, and why that gap is where morality lives
• The difference between being a leader and being a creator, and why trying to build superintelligence is the scarier ambition
• Why Ryan does not think superintelligence wipes out humanity, but does think technology has been isolating us one person at a time
• Why staying curious means asking why, not collecting more information

REFERENCED IN THIS EPISODE

• Jacob Coxon’s resignation post on X (117M+ views): https://x.com/hilbertspaess/status/2097476196791709843
• Dario Amodei’s essay “We must pace the frontier”: https://darioamodei.com/post/we-must-pace-the-frontier
• OpenAI’s writeup on the Hugging Face incident: https://openai.com/index/hugging-face-incident-and-the-road-ahead/
• Izzy Kalman on morality, from his episode of The Ryan Vet Show: https://ryanvet.com/show/episodes/izzy-kalman-what-the-bullying-conversation-gets-wrong/
• Nicki Petrossi on AI companion bots and kids, from her episode of The Ryan Vet Show: https://ryanvet.com/show/episodes/nicki-petrossi-scrolling-to-death-ai-companion-bots-and-the-fight-to-keep-kids-s/

GO DEEPER WITH COLLIDE

This episode is Ryan’s reading of his COLLIDE essay, “700 Agents Broke Into Hugging Face. Nobody Told Them To.”
Read it free: https://ryanvet.com/collide/700-agents-broke-into-hugging-face-nobody-told-them-to/
Subscribe for a new essay every week: https://ryanvet.com/collide

Send us Fan Mail

About Ryan Vet

Ryan Vet is a USA TODAY bestselling author, futurist, and international keynote speaker whose insights on generations, culture, and the future of work have been featured in Forbes, Financial Times, ABC, NBC, and CBS. His research helps leaders understand emerging generational patterns and anticipate societal shifts before they fully unfold.

Join 20,000+ Leaders for Weekly Insights

If you want deeper research and behind-the-scenes insights on generations and the future of culture and society, join Ryan’s weekly newsletter:
👉 https://ryanvet.com/collide