AI Future Fridays: Protecting Your Identity From AI Scammers

Welcome to AI Future Fridays. Every week this column takes one force shaping the near future, tells you the truth about it, and hands you something to build before Monday. First up: the people trying to steal your name, your voice, and your savings, and the surprisingly fun ways to shut them out.

Here's the shift that matters. For more than two centuries, running a con took human labor: someone had to write the letter, work the phone, learn the mark. In the last year that labor cost collapsed to nearly nothing. But the same collapse happened on defense, and that second half of the story is the one almost nobody tells. You can now do things to protect your family that used to require a corporate security budget.

The ID Theft Signal

One phone call shows you the whole future, both halves of it. Dolores is 71, retired in Tucson, and on a Tuesday evening in June her phone rings with her grandson's voice on it. He's crying. There's been a crash, he's in a holding cell, and a lawyer needs $4,300 wired tonight. The voice is perfect: his laugh-swallowed way of saying "Grandma," the flat vowels he got from his dad. Three seconds of a fishing video on his public profile was all it took to build.

Dolores asks one question: "What's the word?"

Silence. Then the line goes dead. Her family picked a safe word at Thanksgiving, a dull little secret that lives in no database, and it just outperformed a million-dollar model. That pause before the click? That's the sound of the future working.

The Backstory of the Nigerian Prince Scam

Every future has a lineage, and this one is older than the telephone. The scam flooding 2026 inboxes is not new crime; it's ancient crime with a collapsing production cost. Which brings us to a myth worth killing: the "Nigerian prince" scam did not come from Nigeria.

The con is European, and it has a paper trail. In the 1790s, prisoners in France ran a scheme the police called the Letters from Jerusalem: a stranger writes claiming to know of a hidden fortune, recoverable if you'll just cover a small fee. Eugène Vidocq, the ex-criminal who went on to found France's civilian police force, watched it run from inside a prison at Bicêtre and reckoned about one letter in five got a reply. A century later American newspapers renamed the same swindle the Spanish Prisoner, a history Atlas Obscura has traced in loving detail. Only in the 1980s, when Nigeria's oil economy cracked and fax machines got cheap, did the old letter put on royal robes and a new return address. The scam got its "419" nickname from a section of the Nigerian criminal code, and an entire country got stereotyped for a French invention that was nearly 200 years old by the time it arrived.

The lineage since then is a story of each new medium inheriting the same old con:

• 1797: Letters from Jerusalem work the Paris post. Reply rate: roughly 20 percent.
• 1898: The Spanish Prisoner works the American mail, one telegraph-era "prince" at a time.
• 1988: The Morris worm knocks out a tenth of the early internet and teaches everyone that networks have locks worth picking.
• 2017: The Equifax breach exposes the personal data of 147 million people, seeding a decade of identity fraud.
• 2025: The shift from generative to functional AI turns one scammer into a thousand personalized, fluent, tireless ones.

Two hundred thirty years, one script: obstructed treasure, urgent fee, borrowed trust. Only the costume changes.

The Forecast for Personal Cybersecurity

The honest read on the next 18 months is that it gets noisier before it gets better. The FBI's internet crime unit counted nearly $900 million in reported losses tied to AI-assisted fraud in 2025 alone, a figure CNN highlighted alongside a wave of voice cloning cases this spring. The share of phishing that shows AI fingerprints jumped fourteen-fold in a single month last winter and has settled at around 40 percent of everything reported, per Tech Times. And researchers told Fortune that voice cloning has crossed the "indistinguishable threshold": human listeners can no longer reliably pick the fake.

Expect the second half of 2026 to be the predators' loudest season yet. The playbook is multi-channel now: a text that moves to an encrypted chat, a follow-up call in a familiar voice, a payment page that looks like your bank's. Data brokers supply the personalization; scraped audio supplies the voice; the con supplies itself, on a schedule, at scale. Businesses feel it too, which is why so many teams are wiring alerts and triage into their own security stack automation rather than watching dashboards by hand.

The con is ancient. The costume is new. What changed in 2026 is that the costumes sew themselves.

The Top Scams to Watch in Late 2026

The Nigerian prince's descendants wear a lot of uniforms now. These are the scripts running hardest right now, and every one of them shares the same three tells: manufactured urgency, an unusual payment rail (gift cards, wire transfers, crypto), and pressure to keep the call secret.

• The jury duty scam. A "sheriff's deputy" or "U.S. Marshal" calls: you missed jury service, there's a warrant, and you can clear it tonight with gift cards or a prepaid card. The federal judiciary's own warning at uscourts.gov is blunt: real courts will never call to demand payment or threaten arrest over the phone.
• The unpaid toll text. A text says you owe a few dollars on a toll or a package redelivery fee, with a link to pay. The amount is small on purpose; the prize is the card number you type in.
• The "safe account" call. Someone from your "bank's fraud team" says your account is compromised and walks you through moving money to a safe account. Banks never ask you to move money. Hang up, call the number on your card.
• Pig butchering. The long game: weeks of friendly texts from a "wrong number" or a dating match, then a can't-miss crypto investment. The romance is the costume; the exit scam is the con.
• The fake job offer. A recruiter, a real-sounding company, a quick hire, then a request to pay for equipment or "onboarding" up front. Real employers pay you.

Grim enough? Good. Take a breath, because every paragraph from here forward is about what you get to do about it, and some of it is fun.

The Countermove Against AI Scammers

The same collapse in cost that armed the scammers just armed you, and this is where the column earns its cyberpunk streak: the street finds its own uses for things. You don't wait for an institution to protect you. You wire the perimeter yourself, in an afternoon, mostly for free, and the ready-made agent workflows people already run are full of the same watch-filter-report shape you'll want.

Start with the boring, load-bearing defenses. They're unglamorous the way a deadbolt is unglamorous:

• A family safe word. Boring, un-Googleable, never written down. Pair it with the callback rule: money never moves on a single incoming call, no matter whose voice is asking.
• A credit freeze at all three bureaus. Free, online, done in an evening. The FTC logged over a million identity theft reports in 2024; a freeze turns stolen data into a key for a lock that no longer exists.
• Passkeys wherever they're offered. No password, nothing to phish, nothing to leak.
• Scam call screening. Android's fake call detection, described on Google's security blog, listens for scam patterns in real time, and a free Google Voice number can make every unknown caller state their business before your phone ever rings.

Next, cut the supply line: get your data out of the brokers' hands. Every tailored scam starts with a dossier someone bought, and the people-search sites are where your address, relatives, and phone number sit in a shop window. There's no universal opt-out yet, so start where it counts: Whitepages, Spokeo, Intelius, BeenVerified, MyLife, TruthFinder, PeopleFinders, and Radaris are the eight that dominate search results. On each one, request deletion AND suppression: deletion alone bounces back when the broker re-ingests public records, usually within three to six months, so suppression is what makes it stick. Californians got a cheat code this year: one request through the state privacy agency's DROP platform now reaches every registered broker in the state, more than 500 of them.

Then build the fun layer: your own watcher. Picture a little agent that runs while you sleep. It sweeps Have I Been Pwned for new breaches touching your email addresses. It reads your inbox the way a bouncer reads a fake ID, flagging look-alike senders and urgency scripts. Once a week it hands you a one-page exposure report with anything that needs your attention. This is real-time threat monitoring, the thing banks pay teams for, running for one household. And it's a pattern anyone can have built.

You describe what the watcher should do in plain English. That description becomes the agent. If you can explain the rule to a smart friend ("flag any email that pressures me to act in under an hour"), you can have BYOBot turn that description into a working agent without writing a line of code.

The Horizon for AI Scamming

Now let yourself see where this goes, because it goes somewhere beautiful. The Letters from Jerusalem survived 230 years because attack always scaled and defense never did: one con letter could hunt ten thousand mailboxes, but every defense had to be taught one grandmother at a time. That asymmetry is dying right now, in our decade. When every phone screens like a switchboard operator, every inbox reads like a bouncer, and every family holds a secret no model can scrape, the economics of the con collapse the way smallpox collapsed: not because predators became kind, but because the herd got immune.

And the defenders hold the one advantage that compounds: we cooperate. Scammers can't share their playbooks without burning them. We can. Every scam pattern one person's watcher catches can teach everyone else's by morning.

One promise this column makes on its first Friday and intends to keep: every future we imagine here puts ending war, halting genocides, and reversing the climate crisis at the top of the list, or else it's a future where those fights are already won. And no future we describe will ever buy safety with somebody else's freedom. No AI wardens, no predictive cages, no watchlists dressed up as protection. All cages are political. The world we're building defends people; it doesn't enclose them. The same pattern-watchers that catch a cloned voice tonight are cousins to the ones learning to spot crop failure, flood risk, and the early signatures of atrocity, and pointing them there is the whole point of getting good at this now.

Dolores hangs up the phone and goes back to her dinner. Multiply her by everyone.

Before Monday: Your Identity Protection Checklist

You have a weekend in front of you. Here's what to do with it:

1. Pick a family safe word at dinner tonight. Boring beats clever. Agree on the callback rule while you're at it.
2. Freeze your credit at Experian, Equifax, and TransUnion. Roughly 20 minutes, total, and it's free.
3. Turn on passkeys for your email and your bank, the two accounts everything else hangs from.
4. Enable scam call screening on your phone, or route unknown callers through a free screening number.
5. Opt out of the big eight people-search sites (Whitepages, Spokeo, Intelius, BeenVerified, MyLife, TruthFinder, PeopleFinders, Radaris). Request deletion and suppression on each. Budget about 30 minutes; Californians can file one DROP request instead.
6. Put a repeating 90-day reminder on your calendar to re-check those opt-outs, since broker data creeps back after each public-records refresh. Better yet, make the re-check your watcher's job.
7. Describe your inbox watcher and have the agent built. Screening messages for scam patterns is exactly the kind of repetitive judgment call worth handing off; it's the same move as any plan to automate repetitive tasks with AI, just pointed at your own front door.

Frequently Asked Questions

• Did the Nigerian prince scam really come from Nigeria?+
  No. The con was invented in Europe. It shows up in French prisons in the 1790s as the Letters from Jerusalem, was renamed the Spanish Prisoner in the American press a century later, and only put on its Nigerian costume in the fax and email era of the 1980s and 90s. The scam is a European invention more than two centuries old; the name describes the disguise, not the origin.
• What is the single best defense against AI voice cloning scams?+
  A family safe word plus a callback rule. Pick a word that's boring and impossible to find online, agree that money never moves on a single call, and always hang up and call the person back on the number you already have. A cloned voice can copy how someone sounds, but it can't know a secret you never wrote down.
• Can I use AI to fight AI scams without being technical?+
  Yes. Scam call screening is built into modern phones, and the watcher-agent pattern needs a plain-English description, not code. Screening messages, sweeping breach databases, and flagging look-alike senders are repetitive pattern-matching chores, the same shape as any plan to automate repetitive tasks with AI, and an agent handles them around the clock.
• How do I find out if my personal data is already exposed?+
  Start with a breach checker like Have I Been Pwned to see which leaks include your email addresses, then freeze your credit so exposed data can't become new accounts. From there, a weekly automated sweep keeps watch going forward; the agent workflow directory has monitoring patterns you can adapt so you find out before a scammer does.