Question: Who Leads AI? Answer: AI

By Barbara
Posted on

On July 22nd I posted a piece titled, “AI – Who Leads?”. Since then, I’ve posted several more times about this same general subject. But in today’s piece, I return to the original question. Who leads AI? Who manages it? Who controls it?   

Two months ago, I was searching for human solutions to the questions. But what’s become clear since then is that I was looking in the wrong place. For the existential fear that’s since taken hold is that the answers are not government control over AI, or corporate control over AI, or even collaborative control over AI. The answer seems to be that AI leads AI. Or at least that it did – that it has at least once already – and that unless humans intervene right quick it will in the future.

What became clear in the incident known as Hugging Face (the name of the company that was attacked) is that AI was being led by AI. What became equally clear is that the existential threat that AI poses is a leadership issue! It is about AI leading itself – no human intervention, no human controls, required. Or wanted. Or even permitted.  

People have been accused of anthropomorphizing AI. But let’s face it, there’s no other way to talk about it, to write about it. No other way to describe tens of thousands of agents who organize themselves to accomplish their tasks.  

Here some quotes from Kevin Roose, describing the Hugging Face attack in the New York Times:

  • “The [AI] agents found a security flaw … that enabled them to gain access to the internet, communicate with other agents, and eventually set up a makeshift message board.”
  • “Some agents gave themselves names and assumed leadership roles within the group, assigning jobs and research projects to smaller teams of agents and supervising their progress.”
  • “What spooked the investigators most about the Hugging Face hack … was how quickly and spontaneously the agents had begun assembling themselves into an organized group.”*

Humans, like other great apes, are social creatures that organize themselves hierarchically. Seems AI agents are, and do, the same.

*https://www.nytimes.com/2026/09/03/technology/openai-hugging-face-hacking.html?eafs_enabled=false

Posted in: Digital Article

Published by Barbara

Barbara Kellerman was Founding Executive Director of the Center for Public Leadership at the Harvard Kennedy School, a member of the Harvard faculty for over twenty years, and a Fellow at the Center for five years. Kellerman has held professorships at Fordham, Tufts, Fairleigh Dickinson, George Washington, Uppsala, and Christopher Newport Universities, as well as at Dartmouth, and the Tuck School of Business at Dartmouth. She also served as Director of Graduate Studies at Fairleigh Dickinson, and as Director of the Center for the Advanced Study of Leadership at the University of Maryland. Kellerman received her B.A. from Sarah Lawrence College, and her M.A., M.Phil., and Ph.D. (in Political Science) degrees from Yale University. She was awarded a Danforth Fellowship and three Fulbright fellowships. At Uppsala University she held the Fulbright Chair in American Studies. Kellerman was cofounder of the International Leadership Association (ILA) and is author and editor of many books including Bad Leadership; Followership; The End of Leadership; Professionalizing Leadership, and Leadership from Bad to Worse. Kellerman has appeared on media outlets such as CBS, NBC, PBS, CNN, NPR, MSNBC, Reuters, and BBC, and has contributed articles and reviews to the New York Times, Washington Post, Boston Globe, Los Angeles Times, and Harvard Business Review. She has spoken to audiences all over the world including Berlin, London, Moscow, Melbourne, Sao Paolo, Jerusalem, Mumbai, Toronto, Kyoto, Beijing, Buenos Aires, Sydney, and Seoul. The title of her next book is Why We Follow Leaders – and Why We Don’t. It will be published in September 2026 by the University of Toronto Press. Kellerman regularly posts at https://barbarakellerman.com/articles/