Help an older relative start using AI, safely

Short answer

Sit down together for about 45 minutes. Install one AI assistant app, such as ChatGPT, from the official app store, set up their own account with 2-step verification (a code as well as the password), try three tasks they care about with them holding the phone, and leave a printed card of rules by it. Check in after a week. The steps and the card are our suggestion.

Why it is worth an afternoon

Many older adults already use AI, and many are unsure about it. In a University of Michigan poll of 2,883 adults aged 50 and over (February 2025), "55% of people age 50 and older said they have used AI technologies that you speak or type to". Only "50% of adults age 50 and older said they feel confident (6% very, 44% somewhat) in their ability to identify when information created by AI is incorrect", and "81% of people age 50 and older would like to learn more about the risks of AI" (National Poll on Healthy Aging, July 2025). Chatbots specifically are less common: Pew found 23% of US adults aged 65 and over had used one, against 66% of those aged 18 to 29 (Pew Research Center, survey of 5,119 adults, February 2026).

Scams are the other reason to do the setup together. The FTC says "Total fraud losses reported by older adults (ages 60 and over) increased about fourfold from 2020 to 2024", "from about $600 million in 2020 to $2.4 billion in 2024" (FTC, December 2025). In 2025, people aged 60 and over filed 201,266 complaints with the FBI's Internet Crime Complaint Center and reported $7.748 billion in losses, up 59% on 2024 (FBI IC3 2025 report).

Before you sit down

Ask in advance what they would like help with. Bring their phone, a way into their email inbox, and a printed copy of the card below.

The session

Step	Time	Check
1. Install the real app	5 min	Installed from the phone's own app store, published by the company that makes it.
2. Create their account	10 min	Their own email, a strong password or passkey, and 2-step verification on.
3. Privacy and comfort settings	10 min	Settings chosen with Privacy settings pass; text size and voice set to suit them.
4. Three first tasks	15 min	They typed or spoke each prompt (their request to the AI) themselves.
5. The card	5 min	Read aloud together and left by the phone.

1. Install the real app. Pick one assistant and stick with it for the first month. Copycat apps exist: in 2023 the security firm Sophos found "multiple apps masquerading as legitimate, ChatGPT-based chatbots to overcharge users" in "both the Google Play and Apple App Store", and warned that "Simply deleting the fleeceware app will not void the subscription" (Sophos, May 2023). For ChatGPT, OpenAI says: "Make sure you are downloading the app published by OpenAI to ensure you have the official and secure version" (OpenAI Help). Our suggestion: open the app store from the home screen, never from a link or advert, check the publisher before you tap Install, and skip any "free trial" that asks for payment.

2. Create their account. Use their own email, so resets and alerts go to them. Save a strong password in their password manager, or add a passkey (AI-written phishing: stop looking for typos, set up passkeys). Then turn on 2-step verification. In ChatGPT, OpenAI's path is settings, then Security and login, then Multi-factor authentication (MFA), with options including an authenticator app (an app that shows sign-in codes), a code "by SMS or WhatsApp", a passkey or a push prompt (OpenAI Help). A text code is the simplest to explain; tell them nobody genuine will ever ask them to read it out.

3. Privacy and comfort settings. Choose the training, history and memory settings with Privacy settings pass. Then raise the text size in the phone's display or accessibility settings, and show them the microphone button, so they can speak instead of type. See Talk, then edit: dictation and voice mode.

4. Three first tasks. Pick three with low stakes and a quick payoff, in their own words (illustrative):

First prompts · v1.0

Explain this letter in plain words. What does it ask me to do, and by when?
<LETTER_TEXT_WITH_NAMES_AND_NUMBERS_REMOVED>

Help me write a short, polite reply to <WHO> saying <WHAT_I_WANT_TO_SAY>.

I'm going to <PLACE> for <DAYS> days and I walk slowly. Suggest a gentle plan.

I have <INGREDIENTS>. Suggest a simple dinner for two.

Replace the following:

<LETTER_TEXT_WITH_NAMES_AND_NUMBERS_REMOVED>: the letter's text, typed or read from a photo, with account and reference numbers taken out (What not to paste).
<WHO> and <WHAT_I_WANT_TO_SAY>: such as my neighbour and thank you for taking in the parcel.
<PLACE>, <DAYS> and <INGREDIENTS>: their own details. A hobby question works too.

After each answer, ask them: "Does that sound right to you?" For a letter about money, health or the law, the answer is a starting point for a call to the sender, not the decision (Health, money and legal questions: green lights and red lines).

5. The card. Read it aloud together, fill in the blanks, and put it where they keep the phone.

Card for the fridge · v1.0

Using <APP_NAME> safely
1. Never type passwords, bank or card numbers, or ID numbers
   (passport, driving licence, National Insurance or Social Security).
2. The real app does not phone me, text me, or ask me to pay anyone.
   Anyone who does is not the app. Hang up and call <HELPER_NAME>.
3. It can be wrong while sounding sure. Check health and money answers
   with my doctor, pharmacist, bank or an official website.
4. Never install another "AI" app from a link or an advert.
5. Nobody genuine asks me to read out a code sent by text.
6. If unsure, stop, and ring <HELPER_NAME> on <HELPER_NUMBER>.

Replace the following:

<APP_NAME>: the one assistant you installed.
<HELPER_NAME> and <HELPER_NUMBER>: you, or whoever is nearest, with a number already saved in their phone.

Rule 2 is our wording. For calls that claim a relative is in trouble, add the call-back card from AI voice scams: hang up and call back on a number you know; we do not repeat it here.

A week later

How to tell it worked

They open the right app on their own and know which one it is.
They can say the card's rules without reading it.
2-step verification is on, and the card is by the phone.

Sources

University of Michigan. National Poll on Healthy Aging: How older adults use and think about AI, July 2025. Survey. Link
Pew Research Center. How opinions and use of AI differ by age, June 2026. Survey. Link
FTC. Annual report to Congress on protecting older adults, press release, December 2025. Complaint data. Link
FBI. Internet Crime Complaint Center 2025 report. Complaint data. Link
Sophos. Fake ChatGPT apps scam users out of thousands of dollars, May 2023. Security firm research. Link
OpenAI Help. How to find and install the ChatGPT Android app, and multi-factor authentication. Vendor. Link · Link

Changes

v1.010 Oct 2026first version

Short answer

Why it is worth an afternoon

Before you sit down

The session

A week later

How to tell it worked

Sources

Changes

Related