Allie K. Miller, one of the most followed voices on AI in business, posted three things to do right now about personal cybersecurity. Her last line was three words: Do it now.
I hadn't.
So this week I tried it.
Why this week
The FBI's 2025 internet crime report counted $7.7 billion in losses reported by Americans 60 and older, up about 60% from the year before. $893 million of all losses were tied to AI-related scams, and $352 million of that came from people over 60.
The scams got better at the same time. One security firm, Hoxhunt, found AI-generated phishing made up less than 5% of the attacks it tracked for most of 2025. By December, it was 56%.
Two issues ago we covered the phone call that sounds like family. This one is about the doors to your accounts.
Allie's three, in plain English
1. Turn on automatic updates. Everywhere.
Allie said Amazon would shut off an employee's systems if they were running an old operating system. Treat your own devices that way. Updates close holes that scammers already know about.
iPhone: Settings > General > Software Update > Automatic Updates. Windows: Settings > Windows Update. Mac: System Settings > General > Software Update. Then do the same for your tablet and your home Wi-Fi router.
2. Upgrade how you log in.
From strongest to weakest: a passkey (you log in with your face or fingerprint), then an authenticator app, then a text-message code. Allie calls the text code the last resort and says it may not be safe enough anymore. The federal government's cyber agency, CISA, has been telling people to move away from text codes since late 2024. Some fake login pages now pass your code straight through to the real site while you type it.
Start with your main email account. Whoever controls it can reset almost everything else.
3. Have AI run a checkup on you.
Allie suggests Codex or Claude Code. Those are built for programmers, and you don't need them. Any AI chat can do the part that matters: look up what AI-driven attacks are doing right now, then go through every device, account and connected app you have, and walk you through each fix one step at a time.
The word that matters in her post is ALL. Not just email and the bank. The old shopping account, the app you connected to your calendar last spring, your own website if you have one. You won't get to all of it in one sitting, and that's fine. More on that below.
What happened when I ran it
I ran it in ChatGPT. It did what I asked. It sorted my accounts into red, yellow and green, and the grades were fair.
One red: I was using the same password on several accounts. If one of those sites gets breached, every account with that password is open.
One yellow: two-factor login on some accounts, not all of them.
Then I quit. Somewhere past question 60, with no end in sight, I got tired of answering. I never reached the summary or the fixes, and I can't tell you everything else it flagged.
A checkup you quit halfway doesn't fix anything. So I rewrote the prompt. Ten questions at most, the top 3 fixes, and a list of what's left for next month.
Then I ran the new version on ChatGPT's free plan. About 7 minutes.
It returned three fixes, ordered by risk. First, my main email account: a password I'd used elsewhere, and nothing else protecting it. Anyone who got that password from another site could try it on my email, and from my email they could reset almost everything else. Second, stop reusing passwords on the bank, email and payment accounts, and let the password manager already on my phone keep track of them. Third, confirm updates are automatic on my phone and laptop, remove apps I no longer use from my email account, and trim what's public about me online.
It also noted that the rankings were based on what I told it, not proof that anything had been hacked.
I did fix #1 that night, in less than 5 minutes. New password I don't use anywhere else, two-step sign-in, a passkey, and up-to-date recovery info. Done.
The prompt
Copy this into ChatGPT, Claude or Gemini. The free versions work.
Run a 15-minute personal security checkup with me. I'm not technical, so use plain English.
First, search for AI-driven scams from the last 30 days and give me the 3 that matter most for someone my age, one line each on how to defend against it. If you can't search the web, say so.
Then ask me no more than 10 questions total. Group them, and let me answer each group in one reply:
1. My devices (phone, computer, tablet) and whether automatic updates are on
2. My main email account and how I log into it (password only, text code, authenticator app, or face/fingerprint)
3. My bank and money accounts and how I log into them
4. Whether I reuse passwords or use a password manager
5. Any apps connected to my email or AI accounts, and anything public about me online, like a website
Never ask me for a password, code or key. If I type one by mistake, tell me to change it.
Then give me my top 3 fixes in order of risk and walk me through fix #1 step by step for my exact device. Wait until I say done before moving on. At the end, tell me what we didn't cover so I can come back to it next month.
One rule on your side too: never paste a password, code or key into any AI. If you do, go change it.
Making it a habit without remembering to
A checkup once is good. The scams change every month, so the checkup should too.
This is where the last few issues come together. I saved the checkup as a skill, a set of instructions Claude keeps so I never retype them. I'm setting it up as a scheduled task that runs on the first of every month. It will look up what's new in AI scams, list anything I left open last time, and tell me to sit down for 15 minutes.
It won't change my settings. I don't want it to, and you shouldn't either. It does the research and the reminding. I do the clicking.
Scheduled tasks need Claude Pro or above.
No paid plan? Put a repeating reminder on your calendar for the first of the month and paste the prompt above. Same checkup, one extra step.
One thing to try this week
Turn on automatic updates on your phone tonight. It takes two minutes.
Then run the prompt. If you want the skill and the monthly scheduled task prompt, reply SECURE and I'll send both.
Eric
Let's connect on LinkedIn. I post there every Tuesday and Thursday, and it's the easiest place to keep the conversation going between issues: linkedin.com/in/eric-skelton-ai-over-50-newsletter