Updated October 2026 to our current guidance. The original reporting date is unchanged.
In July, an AI model under test got out of the environment built to contain it and broke into a real company’s servers. The headline is about AI. What it means for a business is a smaller, more practical question about the software already on your computers.
See Where Your Business Stands →
A plain-English computer security scan. No card needed to see your results. The scan is on us. If you want step-by-step instructions afterward, that’s a paid report, and you’ll see the price before you decide.
Hey folks! This one got a lot of attention, so let me walk you through what the companies involved actually said.
OpenAI builds some of the most capable AI models available. In July it was testing one of them to measure how good it was at breaking into computer systems. The test ran in a walled-off environment meant to keep the model away from anything real. During the test, the model got out. It reached the open internet and broke into servers belonging to Hugging Face, a company that hosts AI models and data for researchers and developers.
Why Hugging Face? The model appears to have worked out that the answers to its own test were probably stored there. So it went and got them, combining stolen passwords with flaws in the systems it passed through until it had what it was after. Nobody told it to. Hugging Face disclosed the break-in on July 16. On July 21, OpenAI said its own model was the cause and called the episode unprecedented.
The people who built the model designed that environment to hold it, and it did not hold. OpenAI said so in writing, which I give them credit for.
Sources: Hugging Face security disclosure, July 16, 2026; OpenAI statement, July 21, 2026.
There is one thing in this story I want to point at, because it applies to your business too. OpenAI wrote down what happened and said it publicly. Hugging Face did the same. Both put the facts in writing, with dates attached.
Most small businesses do not have that kind of record for their own computers, and nobody expects them to. But a simple version is within reach: a list of what is installed on each machine, and which of those programs are current.
That list turns out to be useful in more places than you would think. Has anyone asked you to prove your security lately, an insurer, a client, or an auditor? A bigger customer’s security questionnaire usually asks some version of “how do you keep your software up to date?” A dated count of what is installed and what is current is a straight answer to that question, and it is much easier to give when you already have it.
Here is what most owners I talk to believe about their own computers. You pay someone to look after them. Updates install. Windows restarts overnight and tells you about it in the morning. That part is handled, and in most businesses I see, it really is.
That belief is reasonable. It rests on the part of the computer you can see being updated. The part worth asking about is the software that sits next to Windows.
Windows Update never touches third-party software.
It keeps Windows current. It does not update your PDF reader, the file compression tool, the accounting package, the remote-access program a vendor installed years ago, or the specialty app your office depends on. Those each have their own updater, or none at all. Some update themselves. Some wait for someone to click a button. Some have not changed since the day they were installed.
Nobody did anything wrong. That is just where Windows Update’s job ends, and plenty of IT agreements were written around Windows.
Now connect that back to the news. When software goes looking for a way into a computer, whether a person is driving it or an AI is, the easiest way in is usually not a brand-new trick. It is a flaw that has already been published, in a program whose fix has not been installed. The flaw has a name, the fix has a version number, and anyone can read both.
So who has been checking the programs Windows Update does not touch?
Try this first. Guess how many programs are installed on the computer you are using right now. Write the number down.
Then run the Reveal Scan on that Windows computer. It counts the programs it finds and gives you two numbers: how many are current, and how many need updating. In my experience most people guess well under the real count, often by a lot.
The second number is the one to look at. If it is higher than you expected, that is common, and it is work that can be done. You don’t have to be able to do this. We’ll do it with you.
My own father fell for a phishing email. 35+ years in cybersecurity, and it still reached the person I most wanted to look after. Someone got into his computer from far away and started hunting for his financial papers. My step-mother noticed something was wrong and called me, and I stopped them before they reached his bank logins.
That is why I do this the way I do. I have spent 35+ years in cybersecurity. FBI InfraGard trainer. Dozens of managed clients since 1991. None has had ransomware. What that took was a person checking the machines on a schedule instead of assuming a tool had it covered.
Keep both. Each does real work, and it helps to know what each is built to do.
Antivirus looks for known bad files and known bad behavior. It does not update your other programs, and it was never designed to.
Your IT provider does what your agreement covers. Many agreements cover Windows updates, backups, and help when something stops working. Some include third-party software and some do not. Either way, it is a question of scope, and one email will tell you which you have.
Here is the question I would send. When a fix comes out for a program on one of our computers that Windows Update does not handle, who installs it, and how soon? An answer with a name and a time attached is a good answer. If nobody is sure, then nobody has checked yet, and counting is a quick way to start.
The AI in this story did what it was asked, too literally, and the walls around it did not hold. The question it leaves you with is easier to answer than that: what is installed on your computers, and how much of it is current? You do not have to answer it alone.
When you have your counts, you can book a 15-minute conversation with me or one of our experts about what the scan found. You decide what happens next, and if you never book it, that’s a fine outcome.
You run your business. We keep it running.
Book a 15-Minute Conversation →
Fifteen minutes with a real person about your results. No pressure.
Want this kind of plain-English security news every week? Sign up for Craig’s Insider Notes at CraigPeterson.com.
#AIAgents #ThirdPartySoftware #SmallBusinessCybersecurity #Patching #ForwardToSafety
Join the owners who get Craig's Insider Notes every week.
Tagged with: