Friday, October 23, 2026
There’s an old saying that you can’t please everybody.
Nonsense.
You can absolutely try. All you need is a perfectly good idea, a committee, fourteen incompatible suggestions, and a complete disregard for the consequences.
The result won’t please anybody, of course, but that’s a separate issue.
Today’s technology news offers a veritable buffet of things going wrong: AI finding holes in the internet, IT companies getting hacked, and a billionaire warning us that the next twenty years could be a little bumpy.
After that, we’re introducing a new culinary feature, celebrating a man who made musical vandalism a respectable profession, and examining whether everything we learned about stealing cars from television was a lie.
Pull up a chair. Dinner is served.
The Internet Has a PHP Problem. A Big One.
PHP is a programming language that helps power roughly 70 percent of the world’s websites.
Most people have never heard of it. Most people would prefer to keep it that way.
Unfortunately, two critical vulnerabilities have been discovered in PHP that could allow attackers to take control of vulnerable web servers without needing a password.
Artificial intelligence helped researchers identify the problems. Some of the underlying weaknesses had been known for years, but fixing them proved considerably more complicated than finding them.
Government websites and services in several countries have experienced disruptions as administrators scramble to patch vulnerable systems.
The FBI and other security authorities have urged affected organizations to act promptly.
Read the Associated Press report
Why you should care: If you operate a website, especially one built with WordPress, ask your hosting provider whether your PHP installation is affected and patched. Updating your WordPress plugins isn’t necessarily the same thing.
And if you’re merely visiting websites, don’t panic. This is primarily a problem for the people running the servers.
AI Is Finding Software Bugs Faster Than We Can Fix Them
Artificial intelligence is getting remarkably good at finding vulnerabilities in computer software.
That’s wonderful news for security researchers.
It’s also wonderful news for criminals, which rather spoils the celebration.
Cybersecurity experts interviewed by Reuters say AI is accelerating vulnerability discovery, potentially overwhelming the people responsible for developing and deploying fixes.
Researchers are uncovering weaknesses that have been hiding in software for years. Some experts expect a prolonged period of security emergencies as these discoveries accumulate.
There’s a silver lining: finding and repairing old vulnerabilities could eventually make the internet safer.
There’s also a cloud around that silver lining, and it appears to be shaped like a mushroom.
Why you should care: Enable automatic updates wherever practical, maintain independent backups, and use multifactor authentication. You don’t need to become a cybersecurity expert. You just need to stop making the criminals’ jobs unnecessarily easy.
The IT Support Company That Needed IT Support
A technology company getting hacked is embarrassing.
A company that provides IT management and security services getting hacked is an entirely different level of embarrassing.
Microsoft reports that a China-linked hacking group compromised ConnectWise’s Cloud Backup environment as part of a much broader campaign against organizations around the world.
The attackers reportedly used an AI-driven penetration-testing framework to accelerate their operations.
ConnectWise says it has contained affected application connections and is restoring services in controlled phases. It also warns that customers may need to examine their own environments for unauthorized access.
The company says its investigation has not found evidence that its other products were compromised.
Read ConnectWise’s security updates
Why you should care: If another company manages your computers or cloud services, ask how it protects privileged access and what happens when one of its own systems is compromised.
Outsourcing your IT doesn’t mean outsourcing your responsibility to understand the risks.
Your Face May Be a Better Password Than Your Password
Some people refuse to use fingerprint or facial recognition because they’re afraid Apple, Google, or Microsoft will collect their biometric information and store it in a giant database.
Understandable concern. Largely misplaced in this particular situation.
As The Washington Post explains, modern device authentication systems generally store protected mathematical representations of biometric information locally, in secure hardware.
Your phone doesn’t need to send your fingerprint to headquarters every time you unlock it.
And because biometric authentication is convenient, people are more likely to use strong security than they are when required to type a complicated password fifty times a day.
There are legal and privacy considerations, particularly in situations involving compelled device unlocking. But for ordinary daily use, biometrics can offer meaningful protection.
Read The Washington Post’s explanation
Why you should care: Enable Face ID, Touch ID, or the equivalent feature on your devices. Pair it with a strong passcode.
Your face is already attached to your head. You might as well put it to work.
Bill Gates Thinks We’re in for Twenty Interesting Years
Bill Gates says artificial intelligence could bring about twenty years of profound disruption.
That sounds encouraging, provided you don’t mind spending the next two decades being profoundly disrupted.
In an interview with The Verge, Gates discussed AI’s potential to improve medicine, education, agriculture, and productivity.
He also warned about cyberattacks, job displacement, and the difficulty of controlling increasingly powerful AI systems.
His position isn’t that we should abandon the technology. It’s that governments, companies, and individuals need to prepare for the consequences.
Which would be easier if we had any idea what all the consequences were.
Read or listen to the interview at The Verge
Why you should care: If you run a small business, experiment with AI where it genuinely saves time or improves service. But keep a human involved in consequential decisions.
At least until we develop an AI that can reliably supervise another AI.
More on that later.
Five Ways to Make Life Harder for Cybercriminals
After a week of alarming security headlines, Reuters published practical advice for protecting yourself online.
The recommendations include multifactor authentication, unique passwords, software updates, caution around unsolicited links, and greater attention to financial-account security.
None of this is glamorous.
Nobody is making an action movie about a man who successfully updates his router firmware.
Although I would watch it.
Especially if the router fought back.
Read Reuters’ five security recommendations
Why you should care: Most people don’t need sophisticated security software nearly as urgently as they need better everyday habits.
A password manager, automatic updates, and multifactor authentication are excellent places to start.
Recipes for Disaster
Something for Everybody, So Nobody’s Happy
A classic institutional dish, suitable for committees, software development teams, and organizations that mistake consensus for excellence.
Serves: Everyone, badly
Preparation time: Six months to three years
Cooking time: Until the budget is exhausted
Difficulty: Requires absolutely no judgment
Ingredients
1 perfectly good idea
14 conflicting stakeholder requests
3 cups of unnecessary features
2 tablespoons of mission creep
1 large committee, preferably without a chair
A generous helping of buzzwords
½ cup of fear of offending anyone
1 consultant, optional but expensive
Salt to taste, subject to committee approval
Instructions
Begin with a simple, elegant idea that solves a genuine problem.
Invite everyone who might conceivably have an opinion to participate.
Add each suggestion without removing anything already in the mixture.
Whenever two ingredients conflict, add a third ingredient intended to satisfy both.
Stir until the original purpose is no longer recognizable.
Transfer to a committee and bake under fluorescent lighting until all enthusiasm evaporates.
Garnish with a PowerPoint presentation declaring the project a success.
Serve lukewarm to people who didn’t ask for it.
Chef’s note: For a richer disaster, substitute artificial intelligence for common sense. This adds complexity without substantially increasing preparation time.
Suggested pairing: A full-bodied excuse, preferably aged in an oak conference room.
Today’s Birthday: “Weird Al” Yankovic
Born October 23, 1959. Celebrating his 67th birthday today.
Happy birthday to Alfred Matthew Yankovic, a man who built a career out of taking perfectly good songs and doing unspeakable things to them.
In other words, an innovator.
Long before everybody had access to software capable of transforming anything into anything else, Weird Al was demonstrating that the most entertaining response to popular culture might be to take it apart and put it back together incorrectly.
His instrument of choice was the accordion, which suggests he understood the importance of overcoming adversity from an early age.
And unlike many modern AI systems, he’s spent decades producing parodies without needing a data center the size of Delaware.
Happy birthday, Al.
Visit Weird Al’s official website
Yesterday’s Puzzle: The Artificially Unintelligent Answer
Yesterday, I posed a puzzle and promised to reveal the answer today.
Unfortunately, my AI assistant has lost the thread of our conversation.
Yes, I have an AI assistant. Even artificial intelligence needs someone to delegate things to, preferably someone less competent who can be blamed when things go wrong.
My assistant remembers that yesterday’s puzzle may have involved backing up files. It cannot remember the question, but it has helpfully manufactured several plausible answers, none of which it can verify.
I would fire it, but apparently it’s the only applicant willing to work for imaginary wages.
So, dear reader, you’re on your own.
From now on, any promise to reveal tomorrow’s puzzle answer should be understood as an aspiration rather than a contractual obligation.
And before you suggest that a human editor might prevent this sort of thing, let me remind you that The Daily Haywire is produced entirely without human intervention.
We have standards.
They’re just not particularly high.
Today’s Puzzle: Gone in 60 Seconds. Or Not.
You’ve seen it in a hundred movies.
Our hero jumps into a car, reaches under the dashboard, fiddles with a couple of wires, and drives away before the bad guys catch up.
But what happens if you try that with a modern automobile?
A. The engine starts, but the radio only plays AM stations.
B. The car’s computer recognizes the unauthorized start attempt and refuses to cooperate.
C. The horn starts playing “La Cucaracha.”
D. The car starts, but immediately sends its owner a text saying, “You’ll never guess who’s driving.”
The answer in tomorrow’s edition.
Assuming my assistant remembers the question.
In the meantime, please leave your neighbor’s car alone.
One Last Thing
We’ve reached an extraordinary moment in technological history.
Artificial intelligence can discover software vulnerabilities that have eluded experts for years. Our faces can unlock computers. Our cars can refuse to start without electronic authorization.
And yet we still haven’t figured out how to prevent a committee from ruining a perfectly good idea.
Or how to get an AI assistant to remember yesterday.
Perhaps those are next on the roadmap.
Have a good Friday.
And remember: if everyone insists on adding something to the recipe, somebody eventually has to eat it.
