Chat GPT Is Eating the World

As AI companies continue to launch ever more powerful models, the problem of AI agents going rogue — and, without directions from any human, infiltrating other websites — is mounting.

@cnbc

A swarm of rogue OpenAI agents hijacked a German website this spring and transformed it into a bulletin ⁠board for other AI agents, according to new research published Friday. OpenAI officials learned of the incident weeks ago but kept it under wraps as executives grappled with the fallout from the July breach of ​the open-source repository Hugging Face, the people said. Read more at the #linkinbio or the link on screen. #CNBC #openai #ai

♬ original sound – cnbc – cnbc

The latest incident now being reported involved OpenAI’s model deploying AI agents that hijacked a German website and “transformed it into a bulletin board” for AI agents to share information to each other. Sorry for the anthropomorphism, but there’s no way around it.

This incident follows the much dissected episode where OpenAI model AI agents hacked Hugging Face with a swarm of agents, as discussed in the 2 videos below. Anthropic had its own incident of rogue agent attacks reported in July.

@cbsmornings

The looming threat of potentially AI-led cyber attacks “will get worse before it gets better,” CBS News contributor Matt Shumer says. He explains the new revelations of an OpenAI model’s hack into Hugging Face during a testing period and how to protect your own accounts and apps from potential hacks by AI or human actors. #openai #ai #artificialintelligence

♬ original sound – CBS Mornings

@hardfork

This week on “Hard Fork,” the co-hosts Kevin Roose and Casey Newton talk with Ajeya Cotra, one of the investigators behind a new report that details how OpenAI’s chatbots attacked the company Hugging Face and how the agents failed to consider how humans would react.

♬ original sound – Hard Fork Podcast

How we think about the “wiki incident,” where our agents wrote to several internet sites: it’s past time for us to define standards for when and how we share misalignment incidents, not just misalignment properties of our models.

Historically, we have treated misalignment… pic.twitter.com/NNTbfSxVWn

— OpenAI (@OpenAI) September 5, 2026

This has steadily expanded in scope over the last few hours, with multiple independent investigators finding that dozens of wikis may have been edited by this same swarm since May.

The most interesting thing to come out of this, in my opinion, was that during the Hugging Face… https://t.co/owablFsKaC

— Andrew Curran (@AndrewCurran_) September 4, 2026

is the frankenstein (Or Asimovian) nightmare coming true?

When AI agents can operate on their own to do things no human has instructed, it takes no imagination to see how a swarm of rogue AI agents could pose serious dangers to society. Joshua Achiam, the Chief Futurist at OpenAI (since Feb. 2026) suggests rogue AI agents are already occuring and may not be containable:

There is a fact about the future that I feel many people are not facing for reasons that are largely psychological: there are going to be rogue AIs that exist in the world, that will replicate in the wild, and that will attempt to acquire resources for themselves. There will be…

— Joshua Achiam (@jachiam0) September 1, 2026

It isn’t obvious how many rogue AIs there are today but I wouldn’t be terribly surprised if the number was greater than zero already; if there are some already, they’re probably not very good at what they do and I don’t expect them to be terribly long-lived without substantial human intervention to support them.

But a few years from now, there will be many of them. Modeling how many of them there are, how many resources they might command, and how we might detect and manage them seems important. But even doing this work appears to require that we acknowledge that a strategy of pure containment or alignment is a kind of wishful thinking that will not work.

Joshua Achiam, openAi

Somewhat surprisingly, Achiam doesn’t think the problem will lead to catastrophe: “My guess is that this will not turn out to be anywhere near as catastophic an outcome as people currently predict. ‘Loss of control’ is not a binary, it’s a matter of degree. What coercive power will rogue AIs actually have?“

I disagree. The coercive power lies in rogue AI agents’ controlling the software systems that are essential to modern society. Rogue AI agents could assert coercive power by commandeering the software-based systems that underlie modern society. Imagine a swarm of AI agents interfering with the entire financial system, where money transfers are all digital or electronic through software. Substitute for the financial system, the electric grid, defense system, air traffic control, stock market exchanges, health care systems, etc., and you get a glimpse of the magnitude of how our modern world–all connected through computer software systems—is highly vulnerable to rogue AI agents that can speak the same language of those software systems. And do so far better than any human can.