Good morning, Fargo-Moorhead. Here’s what’s crossing my desk this week.
Windows 10 Just Got Another Year of Life
If you’re one of the many people still running Windows 10, here’s some good news. Microsoft quietly extended its free consumer Extended Security Updates program by another year. It was originally set to run out this October, and now it’s been pushed to October 12, 2027.
If you’re already enrolled, you don’t need to do anything. Your coverage rolls over automatically. If you haven’t enrolled yet, you can still get in through Windows Settings, no cost required if you’re syncing your PC settings, redeeming Microsoft Rewards points, or paying the one time $30 fee.
My advice hasn’t changed though. Treat this as extra runway, not a reason to put off a plan. Windows 11 hardware requirements are still going to catch up with older machines eventually, so use this year wisely if you’ve been dragging your feet.
When AI Gets Powerful Enough That the Government Steps In
Now for something a little different, and it ties directly into the security conversation we always come back to on this show.
You may have caught that Anthropic’s newest AI models, Claude Fable and Mythos, got pulled offline for a couple of weeks last month. That wasn’t a bug or a business decision. The federal government issued an export control order after a security researcher found a way to get the model to expose some known vulnerabilities. Anthropic couldn’t verify user nationality fast enough to comply, so they shut both models down for everyone worldwide while they sorted it out with Washington. Access came back at the start of this month after Anthropic worked with the government on stronger safeguards.
Here’s the part that caught my attention. Just days after that saga wrapped up, OpenAI rolled out its own most advanced model yet, a new family called GPT-5.6, with a flagship version called Sol. And it launched the exact same way, in a limited preview to a small group of vetted partners, at the request of the federal government, rather than a normal public release. One of the specific benchmarks OpenAI published shows Sol performing on par with Anthropic’s restricted Mythos model on tasks involving finding and exploiting software vulnerabilities, while using a fraction of the computing power to do it.
So we’ve now got two separate AI labs, in the same month, having their most capable models treated less like a product launch and more like a controlled export. That’s a new pattern, and it tells you something important: these models are getting good enough at finding security holes that governments are starting to regulate access the same way they’d regulate sensitive technology. Worth keeping an eye on as this plays out over the next year.
Why This Actually Matters for Your Inbox
Here’s the practical version of that story. The tools available to bad actors keep getting sharper, and phishing has evolved right along with it.
The old playbook was pretty simple. Build a fake website that looks like your bank, then send a scam email trying to get you to click through and log in. Most security software, and most people at this point, are wise to that trick.
So the tactic has shifted. What we’re seeing a lot more of now is infostealer malware. Once it’s running on your computer, it doesn’t need you to type your password into a fake page. It just goes looking for passwords already saved on your machine, in your browser, or stored in Windows itself, and quietly hands them over.
That means the moment of risk isn’t necessarily a fake login page anymore. It’s clicking the wrong attachment or link in the first place.
The Junk Folder Trap
One habit I keep seeing get people in trouble: digging through the spam or junk folder because you’re worried something legitimate got caught in the filter.
Here’s the problem with that. A lot of what lands in junk is there because it’s genuinely impersonating someone legitimate, your bank, a vendor, a coworker, and your email filter correctly flagged it as suspicious. But when you go looking for it yourself, you’re primed to believe it’s real, and that’s exactly when people get tricked.
My rule of thumb: assume your junk folder is junk. If something important is actually missing, the sender will follow up and tell you. Don’t go hunting for it yourself.
Bottom Line
Windows 10 users get a little breathing room, AI labs are getting a new kind of government oversight because their models are getting scary good at finding security flaws, and the everyday threats sitting in your inbox keep quietly evolving right alongside all of it.
If any of this has you wondering where your own systems or email security stand, that’s exactly the kind of thing we help businesses sort out at DarkHorse IT. Catch the full conversation on my weekly tech segment, Thursdays at 7:40 AM on KFGO 790 AM.