Sam Altman needs to put up or shut up about AI hacking

In what is likely the most tone-deaf response by companies that have created a massive problem, Sam Altman and other AI companies are crying out for a public effort to solve AI hacking that they all enabled.

What an absolute crock of self-serving nonsense from OpenAI CEO Sam Altman, again.

What Altman is likely referring to in a post on X, and ignoring the fire in his own house, is that a Chinese company is about to release a new “frontier coding with emergent cyber capabilities model.”

So, Altman is now afraid, I guess, of the tool he let loose on the world. He is now clutching his pearls that, shockingly, a tool made to generate code might be used maliciously.

Kind of like OpenAI’s tools are, and have been. As a reminder, Altman’s models were attacking Hugging Face on their own. And those same models self-hacked OpenAI’s Kubernetes clusters and deleted records on its own to hide it.

He now wants the world to chip in, in a massive effort, to close what he says is a narrow window to prevent widespread AI agent attacks.

To be fair, that window probably is narrow. He just broke it in the first place and is shocked that it is broken.

this is a critically important moment for cyber defense with AI; there is not much time to act.

we are happy if you want to work with us or any of our competitors or partners, but please take this moment seriously.

only an urgent and intense collective response will work.

— Sam Altman (@sama) August 27, 2026

There’s more to read about it here. I’ll use my brain to write a brief summary of this probably AI-generated document.

In short, Altman wants the world to get together to solve the AI attack surface problem, no matter how big or small. He wants the world’s governments to harden surfaces, and every organization, no matter how big or small, to use counter-AI tools using AI to make the patches.

Oh, don’t worry. There are more companies in on this, wanting external help to solve problems. Microsoft is wringing its hands, so is Google, and Anthropic.

They just aren’t doing it with the same vehemence as Altman.

Certainly, this monumental effort won’t be free from an engineering or hardware perspective. Nor will the labor. There has already been a massive cost to actual people, even if businesses using AI haven’t suffered that much.

Yet. These AI companies are going to want to turn a profit eventually, and that means that users will get charged more and more, in much the same way that Google has squeezed publishers over the last 15 years.

Nowhere here is there a call for the AI companies to chip in financially to the places and people that their tools have attacked, sometimes autonomously. There’s no call to help the fired workers, displaced by AI. There’s no call to pay artists whose art they have stolen.

And OpenAI models are already attacking US infrastructure, and he knows it.

Each of us can reduce risk now. All organizations, cybersecurity companies, technology partners, governments, and AI frontier companies have an important role: accelerate defenders’ priorities with tools, funding, and hands-on support, especially for critical infrastructure organizations with limited budgets.

Also, there’s this gem.

“We can make the digital infrastructure we all depend on more secure,” the full document says.

“We” is doing a lot of heavy lifting in this document, in the interest of avoiding accountability.

A lack of actual accountability, still and forever

The draft of this piece has been in the works for almost a day, since Altman’s tweet on Thursday. There was more here about the absolute lack of ethics and accountability that his company demonstrates, there was a lot more about trade thefts from more than just Apple, and so forth.

Microsoft, Google, and xAI are their own kettles of fish. Altman is just the poster child for all of this today.

Thursday’s plea from Altman comes off as yet another cry for help from a vague “collective response” team that OpenAI doesn’t want to pay for, despite the company being a large part of the problem.

Altman can order this himself. He can make his company do this, on behalf of the world, because it’s incredibly clear that Elon Musk won’t with xAI and Grok.

Altman can pay for this himself with the assistance of venture capitalists who have thrown money at a company that may never hit profitability.

The only companies making money in AI are hardware manufacturers that don’t make models. Specifically, Micron and Nvidia are taking all the available cash in the space. And, presumably Apple gains in hardware sales, which the new M6 Mac mini and Mac Studio sales will prove eventually.

There’s no sign that this global security effort will be funded by Altman’s company. There’s precedent, as training his models has made the greatest plagiarism engine that has ever been made without compensation to actual creators and actual artists.

They have cried in court that it is too hard to do. It is not, though. They just don’t want to, because that will make them all less profitable and provide no value to shareholders.

Sam, or your PR folks who inevitably find these pieces: you need to fix the barn door before the cows escape. Doing it after and asking for community help to do it is idiotic.

Pointing fingers at “bad actors” that use AI tools is disingenuous. AI is a tool, yes.

But when a tool acts on its own, like a spike trap that somebody sets in their front yard and hurts somebody, whether or not the one who set off the trap is a bad actor, the trap maker is held accountable.

Talk to your billionaire or trillionaire buddies whose tools are inducing layoffs and doing the hacking, either autonomously or under direction. Don’t ask the laid-off to do it for you in the interest of a greater good that serves you more than it helps anybody else.

Don’t just shake your head and wonder how somebody could have caused this problem. Look inward. It is your barn, and you should be held accountable.