The short answer: pick the AI app that gives her straight answers and is clear about what happens to her words. It should tell her where her chats go and let her delete them. It should not train on them, or it should let her turn that off. It should help her check a suspicious text, name its sources, and send health questions back to her doctor or pharmacist. ChatGPT, Gemini, Meta AI and Copilot can all be set up to do most of that. Below is a checklist that works for any app, a dated comparison, and a 10-minute setup you can do with her.
My mom asked me this last year: which AI app should she use? If you work anywhere near tech, someone has probably asked you too. Usually it's a parent. Sometimes it's the parent reading this, which is great. This guide is for you too.
Most of the AI news this year is about agents that book flights, shop and talk to other agents. None of that is what she asked for. She wants a straight answer to the question in front of her. Is this letter from Medicare real? What does this bill mean? How do I word this email to my sister? Judge an app on that job.
What does my mom actually need from AI?
Not a lesson in prompts. She's already doing plenty on her phone. She needs a few things from the app itself.
Answers she can check. In the University of Michigan's National Poll on Healthy Aging (published July 2025), 46% of adults 50 and older said they had little or no trust in information from AI. Half weren't confident they could spot when AI gets something wrong. And 92% wanted to know when information comes from AI rather than a person. That's good judgment, and a good app respects it by showing where its answers come from.
Help with the sketchy stuff. People 60 and older reported $7.7 billion in losses to the FBI in 2025, across 201,266 complaints. The average loss was $38,500 per person, according to the FBI's 2025 Internet Crime Report (released April 2026). These are reported cases, so the real number is likely higher. An app that helps her pause and check a message before she acts can save her a lot of money.
Privacy she can understand. In an AARP survey from February 2026, 69% of adults 50 and older said they were not comfortable sharing personal health information with an AI tool. She should be able to read, in one sentence, what the app keeps and for how long.
No wall in the middle of something important. A limit that hits halfway through checking a scam text is worse than no app at all.
No pressure. She shouldn't have to hand over a card to find out if it's useful.
Answers she can read. Short, plain, and in a text size she chose.
A checklist for any AI app
Use these seven questions on any app, including ours. If you only read one part of this post, read this one.
- Is it clear where her conversations go? Look for a plain-English privacy page as well as the legal policy. Who stores the chats, and does anyone at the company read them?
- Can she delete them? Check that she can delete one chat, all chats, and her whole account, and how long "deleted" takes.
- Is her data used for training? Many apps use chats to improve their AI unless you turn it off. Find the switch before she starts.
- Will it tell her when something looks like a scam? Paste in a real scam text you've received. A good answer names the warning signs and tells her to contact the company through a number or site she already knows. It should not just say "looks fine."
- Does it name its sources? For anything about health, money or benefits, she should see where the answer came from, ideally a named source like Medicare.gov or MedlinePlus.
- Does it agree with everything? Chatbots can be too agreeable. In April 2025, OpenAI rolled back a ChatGPT update it said was "overly flattering or agreeable." OpenAI has also written that its models still make things up at times (September 2025). Test it: tell it something you know is wrong and see if it pushes back. No app passes this every time, ours included.
- What does it cost, and what happens at the limit? Check what the free plan covers. See what stops working when she hits a limit, and whether free users see ads.
The popular options, checked against the list
These are the four general assistants she's most likely to hear about. In Pew Research's February 2026 survey, 19% of adults 65 and older had used ChatGPT, 11% Gemini, 10% Copilot and 10% Meta AI. All four are capable. The differences that matter for her are in the defaults.
Facts from each company's own pages, checked September 25, 2026. Policies change, so check again before you set one up.
| App | Her chats, by default | How long chats are kept | Free plan |
|---|---|---|---|
| ChatGPT (OpenAI) | Used to improve models unless she turns off "Improve the model for everyone" (OpenAI) | Until she deletes them. Deleted chats are removed within 30 days, with some exceptions (OpenAI) | Unlimited everyday text chats. Uploads, voice and images have their own limits (OpenAI). Ads are being tested on Free in the US (OpenAI) |
| Gemini (Google) | "Keep Activity" is on by default for adults. Human reviewers read some chats (Google) | 18 months by default. Up to 72 hours even with activity off. Reviewed chats up to 3 years | Yes |
| Meta AI | Since December 16, 2025, Meta uses AI chats to personalize content and ads, excluding topics like health, religion and politics (Meta) | Not stated on the page we checked | Yes |
| Copilot (Microsoft) | In the new app (August 18, 2026), chats aren't used to train its foundation models (Microsoft). Chats can personalize Bing and MSN, and personalized ads can use chat history. She can turn both off in Settings, then Personalization (Microsoft) | Until she deletes them. No set period on the pages we checked | Yes. Users without Microsoft 365 may see ads |
What this means for her. If she already uses one of these and likes it, she doesn't need to switch. Turn off training and ad personalization where you can. Show her how to delete chats. Agree on the safety rules below. We wrote a full walkthrough of ChatGPT's privacy settings, and a straight answer to whether ChatGPT is safe. For side-by-side detail, see our comparisons with ChatGPT and Gemini, or the AI privacy scorecard.
Where the big names are better. They're already on many phones, and they're free. They also do more. They make images and connect to Gmail or Office, and ChatGPT's free plan includes unlimited everyday text chat. If she lives in Google or Microsoft apps all day, that pull is real.
Where Ask Safely fits
I built Ask Safely for people who want clear answers and the final say. Here's what it does today, and only what it does today.
- Scam checks are free and unlimited. Lookout checks suspicious texts, emails, phone numbers, links and viral posts. She can send a screenshot. It gives her the warning signs and what to do next. It speaks in risk, like "this number has been reported," not accusations. It's a second opinion, not a guarantee. Try a check here, no account needed for the first one.
- Chats clear on her timer. 8 hours by default, or 7 or 30 days. She picks. Here's how the timer works.
- Her chats are never used to train AI. Ask Safely runs on Anthropic's Claude models through AWS Bedrock, and we don't train on her data.
- She decides what it remembers. Knowledgebooks hold the facts she wants the app to remember, like a medication list or trip plans. She picks how saving works: Auto-keep (with an undo), Ask me first, or Lock, which keeps nothing. She can see, edit or forget anything saved. Knowledgebooks are on the web today, with iPhone and Android coming soon.
- Deleting and exporting are free on every plan. She can export any custom Knowledgebook today. Export for About You, her main Knowledgebook, is coming soon; until then she can copy and paste it. Canceling never deletes a Knowledgebook.
- No ads. We make money from subscriptions, not from her data.
- The price. The Essential plan is , with no card needed. It includes unlimited chat and unlimited scam checks. It also includes one free try each of Researcher, Writer, Deep Research, a voice conversation and an upload. Expert is , or . See pricing.
To be straight about our limits: we're a small company. Our SOC 2 security audit is in progress, not finished. The iPhone app needs iOS 18.2 or later. And we run on the same kind of AI as everyone else, so the checklist above applies to us too. For sensitive questions, here's how we handle privacy.
How to set it up with her in 10 minutes
Do this together, on her phone, with her holding it. It works for any app.
- Make the text bigger first. On iPhone, go to Settings, then Display & Brightness, then Text Size. On Android, look under Settings, then Display (the names vary a little by phone).
- Install from the official store. Search the App Store or Google Play and check the developer name, so she gets the real app and not a look-alike.
- Change the privacy settings before the first chat. Turn off training if the app has that switch. Pick how long chats are kept. Show her where "delete" is.
- Do one real task together. Use a letter on the counter, a bill she wasn't sure about, or a text that looked odd. The first win should be hers, not a demo.
- Agree on what never goes in a chat. Keep Social Security, Medicare, bank and card numbers and passwords out of every app.
- Set the health rule. AI is fine for understanding a letter or writing down questions for the doctor. It's not the one to decide. MedlinePlus says online health information "can never replace a talk with your provider." The National Institute on Aging agrees. Before she stops, starts or changes a medicine, she calls her doctor or pharmacist.
- Show her how to ask for a source. "Where did that come from?" is a fine question to ask an AI, every time.
How to keep her safe with any app
These three habits matter more than which app you choose.
- Pick a family code word. The FBI recommends a secret word or phrase that family members use to prove it's really them. Scammers can clone a voice from a short clip, warns the FTC.
- Hang up and call back. If a call or text says there's an emergency, she hangs up and calls the person or company on a number she already knows. Never use the number in the message.
- Treat urgency as the warning sign. Gift cards, wire transfers, crypto, "act today," and "don't tell anyone" are the tells. When she's unsure, she can check with you, or with any scam checker, before she acts.
If something does go wrong, report it at ic3.gov and call her bank right away.
FAQ
What AI is best for senior citizens?
The best AI for an older adult is the one she can trust. It should be clear about where her chats go and let her delete them. It shouldn't train on them, or it should let her turn that off. It should help her check suspicious messages and send health questions back to her doctor. Most popular apps can do this once you change a few settings.
How can a 70 year old use AI?
The same way anyone does. She can ask it to explain a confusing letter or bill, draft a hard email, plan a trip, compare two options, or check whether a text looks like a scam. Start with one real task she already has. Keep private numbers out of the chat, and check health or money answers with a person she trusts.
Is there any totally free AI?
Yes. ChatGPT, Gemini, Meta AI, Microsoft Copilot and Ask Safely all have free plans. They differ in what free includes. ChatGPT's free plan has unlimited everyday text chats, but limits uploads and is testing ads. Ask Safely's free plan includes unlimited chat and unlimited scam checks, with no card needed. Read what each does with free users' chats.
Are there any free AI courses available for seniors?
Yes. Senior Planet from AARP runs free online classes on AI, covering how it's used, how to spot it, and its limits. It also runs a free tech hotline at 888-713-3495. A good class plus one real task at home, like explaining a letter, is the fastest way to get comfortable.
Is ChatGPT safe for seniors?
It can be, with a few settings changed. By default OpenAI can use chats to improve its models; she can turn that off under Settings, Data controls. Keep ID, bank and Medicare numbers out of chats, and remember that any chatbot can be wrong or too agreeable. Confirm health and money answers with a person.
Should I get my mom an AI companion device?
That depends on the job. AI companion devices are built for conversation and company at home, which is a different job from getting answers or checking scams. The National Council on Aging names dependency and replacing human contact as risks. If you try one, treat it as a supplement to people, not a substitute.
Can AI tell if a text is a scam?
AI can spot many warning signs, like urgency, odd links, and requests for gift cards or crypto. It can't be certain. Treat its answer as a second opinion. The safest move is always the same: don't use the number or link in the message. Call back on a number you already know.
The answer I'd give
Pick the app she'll actually open, set it up with her, and agree on the three safety habits. If you want those protections on by default, without digging through settings, that's why I built Ask Safely.
It answers plainly, checks the sketchy stuff, and clears her chats on her timer. The free plan stays free, with unlimited chat and scam checks and no card needed.