AI Safety for Seniors

Plain answers about AI. No hype, no fear.

Built for our class at the Senior Center. Everything here is written in plain English and checked against trusted sources.

๐Ÿ“… What's All This Artificial Intelligence About? ยท Tuesdays, Oct 6โ€“27 ยท 1:00โ€“2:00 PM ยท Senior Center Course 542716 ยท $15 ยท Register online or at the Senior Center front desk In the class? Everything you need is at safeai4seniors.com/start โ†’
Safe to use
Use with care
Stop โ€” likely a scam
Scam Spotlight

The "grandchild voice" phone call

Scammers can now copy a family member's voice from a short video clip and call you sounding exactly like them, asking for money urgently. The voice is fake. The panic is the trick.

The rule: Hang up and call your family member back at the number you already have. A real emergency survives a two-minute callback.

๐ŸŽฎ Scam or Safe? โ€” The Game

15 real-looking messages โ€” calls, texts, emails, even an AI chatbot. Can you tell which ones bite? Play one a day, or take on all 15. Great to play with family.

Play Scam or Safe?

๐Ÿง  New: How AI Actually Works

Press Play and watch a ChatGPT-style AI build an answer โ€” one guessed word at a time. Two minutes, no jargon, and suddenly the "magic" makes sense.

Watch how AI works

๐Ÿ“– Class Materials

October 2026 class: slides, handouts and the workbook pages, posted after each Tuesday. The August 2026 materials are kept below as an archive.

๐Ÿšจ Known Scams

The five AI-powered scams hitting people our age right now โ€” real cases, and the simple defense that beats each one.

๐Ÿ” Check a Scam

Got a strange call, email, or text? Answer five quick questions before you click or reply.

๐Ÿ”— Trusted Links

Hand-picked resources: FTC fraud alerts, AARP Fraud Watch, and our San Mateo County helplines.

๐Ÿ“– Class Materials

Our 4-week class: What's All This Artificial Intelligence About? โ€” Tuesdays, October 6โ€“27, 2026. One AI tool a week, met the same way each time. Everything is posted here after each class โ€” click any ๐Ÿ“„ link to open it, then print or save it. Printed copies are always available in class โ€” you never need a computer to keep up.

Questions between classes? Email me any time: rsellers@sonic.net โ€” I read every message.

Start here

๐Ÿš€ The Start Here page

Open this week's tool, flip the privacy switch, the one rule, the three moves, and this week's handouts โ€” in the order you need them.

Go to safeai4seniors.com/start

Current class โ€” October 2026 Now running

Week 1 ยท Tuesday, October 6

Get it in your hand Tool of the week: Gemini

The clone clip, four real stories of people bringing their own material to life, the top 20 things people over 50 use AI for, why experience is an advantage, and a very good guesser โ€” then Gemini on your own phone, three ways: type, talk, show.

๐Ÿ“„ Week 1 Slides ยท ๐Ÿ“„ Take-home: the top 20, the stories, the research ยท ๐Ÿ“„ Gemini Tool Card ยท ๐Ÿ“„ Workbook packet (Week 1)

Homework: use it for one real thing this week. Write down what you asked, what it got right, and what you checked โ€” and where. Bring the page back.

Week 2 ยท Tuesday, October 13

Trust it for words. Check it for facts. Tool of the week: ChatGPT

The second question is where the value is. Your first cross-check: the same question to Gemini and ChatGPT, side by side. Then the level-up โ€” Gemini Notebook (the tool formerly called NotebookLM), an AI that answers only from the documents you give it and shows its source.

๐Ÿ“„ Week 2 Slides ยท ๐Ÿ“„ Take-home: the second question, words vs. facts, the fine print ยท ๐Ÿ“„ ChatGPT Tool Card ยท ๐Ÿ“„ Workbook packet (Week 2)

Homework: one real task with AI, using at least one follow-up. One fact you checked โ€” and where. Try the Notebook if you're ready. Bring the page back.

Week 3 ยท Tuesday, October 20

Safety is a skill, not a fear Tool of the week: Claude

A four-move procedure for anything urgent, the one defence that beats voice cloning (set up in class), and the privacy choices each tool asks you to make.

Posted after class on October 20

Week 4 ยท Tuesday, October 27

Things your kids don't know AI can do yet All three tools

Ask all three, compare โ€” then build something: a family page, a podcast from your own documents, a plan with sources. Each pair presents. Your personal AI plan, and how to keep going.

Posted after class on October 27

Previous classes โ€” August 2026 (archive)

These are the materials from the first run of the course, "AI and You," August 4โ€“25, 2026. They're kept here for the people who took it. Some facts and app settings have changed since then โ€” where the two disagree, the October materials above are current.

August 2026 archive ยท For the whole course

๐Ÿ“„ Take-home guides & homework August 2026

๐Ÿ“„ Student Handout Packet โ€” all 8 take-home guides, from AI Basics to your Personal AI Safety Plan.

๐Ÿ“„ Homework Assignments โ€” all four weekly assignments on one page each.

๐Ÿ“„ Current Topics โ€” Summer 2026 โ€” real Bay Area AI scam stories discussed in class.

August 2026 archive ยท Week 1 ยท Tuesday, August 4

AI Without the Hype August 2026

What AI is (and isn't), in plain terms. Where you already use it every day โ€” voice assistants, fraud detection at your bank, and more. The class mantra: useful AI, minimal exposure, human verification.

๐Ÿ“„ Week 1 Slides ยท ๐Ÿ“„ Welcome Sheet ยท ๐Ÿ“„ Bonus deck: How AI Actually Works ยท ๐Ÿง  The two-minute animated tour

August 2026 archive ยท Week 2 ยท Tuesday, August 11

Staying Safe from AI Scams August 2026

The four shields: pause ยท call back on a number you already have ยท the family safe word ยท never gift cards, wire or crypto.

๐Ÿ“„ Week 2 Slides ยท ๐ŸŽฎ Scam or Safe? โ€” the game

August 2026 archive ยท Week 3 ยท Tuesday, August 18

The Good Side โ€” practical uses August 2026

"Trust it for words. Check it for facts." Letters, paperwork, planning, learning โ€” and how to keep your private information private.

๐Ÿ“„ Week 3 Slides

August 2026 archive ยท Week 4 ยท Tuesday, August 25

Your Personal AI Safety Plan August 2026

Putting it together: your family code word, your callback rule, your trusted-numbers card, and a one-page plan to take home.

๐Ÿ“„ Week 4 Slides ยท ๐Ÿ“„ Trusted Resources one-pager

Missed a class? Materials stay posted here all season, and printed copies are at the Senior Center front desk. Questions? Email rsellers@sonic.net.

๐Ÿšจ Known Scams

These are the scams actually hitting people our age right now โ€” and how AI has made each one more convincing. Every story below really happened. Every defense below really works.

Why this matters: last year, Americans over 60 reported $7.75 billion in fraud losses to the FBI โ€” up 59% in a single year, averaging $38,500 per victim. More than 22,000 of those reports specifically involved AI, with losses of about $893 million. The good news: the five patterns below cover most of it, and each one loses to a simple habit.
Scam #1 ยท The Phone Call

๐Ÿ“ž The cloned-voice family emergency

AI can copy anyone's voice from just a few seconds of audio โ€” a voicemail greeting, or a short video of your grandchild on Facebook. Scammers call sounding exactly like a loved one in trouble: arrested, in the hospital, needing money quietly and immediately. The voice is fake. The panic is the trick.

It really happens: prosecutors charged a ring that ran this scam from an overseas call center, using family details to steal more than $5 million from over 400 victims โ€” average age 84.

The defense: hang up and call your family member back at the number you already have. A real emergency survives a two-minute callback. A family code word makes you even safer.

Where this comes from: FBI ยท Forbes on AI voice-scam losses
Scam #2 ยท The Miracle Cure

๐Ÿ’Š The famous face selling a health "breakthrough"

AI "deepfake" videos put real celebrities' faces and voices on products they've never heard of. Fake videos of Bill Gates, Anderson Cooper, and Bruce Willis have "endorsed" a honey-based Alzheimer's cure. A fake Oprah pushed a weight-loss salt. One woman was fooled by an ad using a deepfake of her own real doctor.

The product is worthless โ€” but the deeper trap is the subscription: recurring credit-card charges victims report being unable to cancel or refund. Anti-aging cures, memory pills, miracle supplements: this is one of the biggest scam categories aimed at people our age.

The honest truth: real healthy aging โ€” movement, sleep, connection, taking care of yourself โ€” is not sold in a bottle from a social-media ad, and it never asks for your card number.

The defense: celebrities and doctors do not sell miracle cures in online ads. Ever. Before buying any health product you saw in an ad, talk to your actual doctor or pharmacist first.

Scam #3 ยท The Email & Text

โœ‰๏ธ The perfect phishing message

You may have learned to spot scam emails by their typos and odd grammar. That advice is out of date: AI now writes most phishing messages, and they arrive flawless โ€” correct logos, your real name, sometimes your real bank. The polish proves nothing anymore.

The defense: judge the ask, not the spelling. Any message that wants you to click a link, "verify" your information, or act urgently gets checked through the front door โ€” type your bank's address yourself, or call the number on the back of your card.

Scam #4 ยท The Official Call

๐Ÿ›๏ธ The government agency that isn't

Government impersonation cost Americans about $920 million last year. The current favorites: a call or official-looking mailer saying Medicare is issuing "new chip-enabled cards" and needs your Medicare number to activate yours โ€” or an AI-generated "official" voice warning your Social Security is about to be suspended unless you act now.

Remember: Medicare and Social Security never call to demand information or payment, and benefits are never suspended by phone.

The defense: hang up. If you're worried, call the number printed on your actual Medicare card or Social Security statement โ€” not any number the caller gives you.

Scam #5 ยท The New Friend

๐Ÿ’” The patient online companion

An attentive stranger appears โ€” on a dating site, on Facebook, or through a "wrong number" text that turns friendly. They are warm, consistent, and never in a hurry. AI supplies their photos (which beat reverse-image search) and can even hold the conversation for them, so one scammer can run dozens of "relationships" at once. After weeks or months comes the crisis ("medical emergency") or the opportunity ("my uncle's investment platform").

The scale: people over 60 lost $584 million to romance and confidence scams last year. Some scammers now even appear on video calls wearing an AI-generated face.

The defense: two unbreakable rules. Never send money to someone you haven't met in person. And anyone who mixes affection with an investment tip is reading from a script โ€” no exceptions.

Where this comes from: FinanceBuzz ยท National Council on Aging
The one thing every scam has in common

They all rush you

Every scam above works the same way: it manufactures urgency or strong emotion so you act before you verify. The panic is the trick โ€” and the pause is the defense.

When anything feels urgent, that's your signal to slow down. Take a breath, and run it through the 5-question check.

๐Ÿ” Check a Scam

Got a call, text, email, or pop-up that feels off? Answer these five questions honestly. Then read your result โ€” before you click, reply, or pay anything.

The 5-Question Check

Tap Yes or No for each one.

1. Did they contact you first โ€” out of the blue?

2. Are they rushing you? ("Act now," "today only," "don't hang up")

3. Are they asking for money, gift cards, wire transfer, or crypto?

4. Are they asking for passwords, codes, Social Security, or bank numbers?

5. Did they say to keep it secret from family or your bank?

Remember the big three

Urgency โ€” real organizations never demand action in minutes. Scammers always do.
Unusual payment โ€” nobody legitimate asks for gift cards, wire transfers, or crypto. Ever.
Secrecy โ€” "don't tell your family" is the surest sign it's a scam. Always tell someone.

The callback rule: hang up, look up the real number yourself, and call back. A real emergency survives a two-minute callback.