How to Avoid AI Scams

For years, we have been told that scams usually give themselves away. Misspelled words, awkward grammar, strange email addresses, or a caller who doesn't quite sound like the person he claims to be are sure signs you should abandon the call or delete and block the email.

Artificial intelligence is making some of those clues less dependable.

Scammers can now use AI to write polished messages, imitate voices, create convincing photographs and videos, and build online identities that look surprisingly real. In 2025, the FBI received more than 22,000 complaints involving AI, with reported losses of more than $893 million. The agency says criminals are using the technology to create fake social profiles, personalized conversations, voice clones, identification documents, and believable video.

This doesn't mean you need to become an expert at spotting deepfakes or figuring out whether every email was written by AI. It does mean we need to change one habit: question the first impression.

Something that looks right, sounds right, or appears to come from someone you know deserves a second look if it asks you to send money, share information, click a link, sign into an account, or do something you weren't expecting.

And this is one place where AI can actually help. Instead of simply asking yourself, Does this look real?, you can give an AI assistant the non-sensitive parts of a suspicious message and ask it to look for problems you may have missed, explain what a legitimate organization would normally do, and help you find a safer way to verify the request.

The Voice on the Phone May Sound Completely Real

A scam that has been around for years is "The Grandparent Scam." Someone calls pretending to be a grandchild who has been arrested, injured, or involved in an accident and needs money immediately. In the past, an unfamiliar voice could give the scam away, but AI voice cloning has made this much harder.

AI can now imitate someone you know by pulling audio from social media posts to create convincing copies of a person's voice. These scams are also expanding beyond grandchildren to other relatives and close friends.

What this might sound like:

Your daughter calls you from an unfamiliar number. She sounds upset and says she has been in an accident. Before you have time to ask many questions, someone claiming to be an attorney gets on the phone and tells you money is needed right away.

Because the voice sounded like your daughter, you may feel you have already confirmed who called.

The safest response, however, is to hang up and call your daughter at the number you normally use. If you cannot reach her, call someone else who would know what is happening.

AI can help too. Once you are off the call, tell your AI assistant what happened and ask it to help you evaluate the situation:

Try this prompt:

Someone called from an unfamiliar number using my daughter's voice and said she had been in an accident. An attorney then asked me to send money immediately. What signs of a family-emergency scam should I look for, and what should I verify before doing anything?

AI cannot tell you whether the caller really was your daughter, but it can recognize the pattern, point out the urgency and payment request, and remind you of ways to verify the story independently.

This distinction is important throughout this article: AI can help you investigate the claim. It cannot replace verification.

A Message Can Look Right and Still Be Wrong

AI can make a fraudulent message look polished, personal, and surprisingly believable. A scammer no longer has to rely on bad grammar or an obviously fake email address. The message may use the right logo, know your name, and sound much like the company or agency it claims to represent.

The details vary, but the pattern is often the same.

What this could look like:

Your bank: You receive an alert saying unusual activity has been detected and you need to click a link immediately to keep your account from being frozen.

A toll agency: A text says you have an unpaid toll and additional penalties will be added unless you pay today.

A utility company: You are told your electricity will be disconnected within the hour unless you make an immediate payment.

A government agency: Someone says your money is at risk and instructs you to move it to a special "safe" account.

A delivery company: A text says your package cannot be delivered until you confirm your address or pay a small fee.

Each message gives you a different reason to act, but the safest response is the same thing: do not use the link, phone number, payment instructions, or contact information supplied in the unexpected message.

Instead, go around it.

Open the bank's app yourself. Type the toll agency's official website into your browser. Find the utility company's number on a bill. Go directly to the government agency's website. Check your package through the retailer or delivery company's app rather than through the text.

If the problem is real, you should be able to find it through a source you chose yourself. AI can help before you take that step. Copy the wording of the message, removing any sensitive information, and ask:

Try this prompt:

I received this message unexpectedly. Review it for signs of phishing, impersonation, or fraud. Tell me what about the request is unusual, what information I should not provide, and how I can independently verify the claim without using any links, phone numbers, or instructions contained in the message.

This is where AI is particularly useful. It can flag urgency, threats, requests for login information, unusual payment methods, or instructions that do not align with how a legitimate organization typically operates.

It may also help you ask a question you would not have thought to ask yourself: Would a bank really handle this via text? Would a government agency ask me to move money? Would a toll authority threaten immediate suspension this way?

AI can help you question the message. The final confirmation should still come from the organization itself.

An Invitation from Someone You Know May Still Be Fake

A recent phishing scam is a good example of why questioning the first impression has become so important.

The FTC has warned about unexpected digital invitations that appear to come from familiar services such as Evite or Paperless Post. Some list someone you actually know as the host. The problem occurs when you try to open the invitation or RSVP and are prompted to enter your email password, phone number, or a verification code.

What this might look like

An invitation to a friend's birthday party arrives by email. You recognize the friend's name and the invitation service, but when you click to RSVP, you are asked to enter your email username and password.

Before entering anything, stop.

You could take a screenshot of the invitation and ask your AI assistant:

Try this prompt:

I received this invitation unexpectedly. Look at it for signs of phishing. Tell me what information a legitimate online invitation should reasonably request, what I should not provide, and how I can verify the invitation's authenticity without using the link.

One detail AI may flag is the request for a verification code. Those short codes can seem harmless, but they may be exactly what a scammer needs to reset a password or gain access to an account.

There is another wrinkle worth knowing. The FTC says that if scammers take over someone's email account, they may use it to send the same fraudulent invitation to that person's contacts. This means the message really could come from your friend's email account and still be fraudulent.

The simplest check has nothing to do with technology: text or call your friend separately and ask whether the invitation came from them.

What About Photographs and Video?

We are accustomed to treating photographs and video as evidence. A detailed photograph makes an online seller seem more credible. Vacation pictures make someone we met online seem more real. A video showing a familiar person speaking feels much harder to question than an email. But, AI is weakening all of those assumptions.

The FBI says criminals are using AI to create fake social profiles, identification documents, voice clones, and believable videos. It also warns that AI-generated content has advanced to the point that it may be difficult to identify fakes simply by looking or listening.

What this might look like:

You are considering buying an expensive item from someone online. The seller sends several detailed photographs and even a short video of the item. Everything appears genuine, so sending a deposit begins to feel reasonably safe.

Before you send any funds, ask AI to examine the images for obvious inconsistencies. But AI-generated images continue to improve, and neither you nor an AI assistant can be expected to catch every fake.

A more useful prompt would be:

Try this prompt:

I'm considering buying an expensive item from someone I don't know online. The seller has provided photographs and video. What should I independently verify before sending money, and what evidence would be more reliable than the images alone?

This shifts the question away from “Can you prove this photograph is fake?” and toward “What else would make this transaction trustworthy?”

If You Think You've Been Scammed

If you discover that you have sent money or shared information with a scammer, act quickly rather than waiting to see what happens.

Contact your bank, credit-card company, or the service you used to send the payment and explain that the transaction may be fraudulent. Ask whether it can be stopped or reversed. If you shared a password, change it immediately anywhere you have reused it and turn on two-factor authentication where available.

If a scammer may have gained access to an email account, change the password promptly. The FTC also recommends going to IdentityTheft.gov for steps based on the information that may have been compromised.

Fraud can be reported to the Federal Trade Commission at ReportFraud.ftc.gov and internet-related crime to the FBI's Internet Crime Complaint Center at IC3.gov. The FBI recommends keeping information about how you were contacted, when the contact occurred, how any payment was made, and where the money was sent.

---

Continue Learning

This article is part of our growing AI for Homeowners Resource Center, where you'll find practical guides, AI tool reviews, homeowner checklists, and prompt libraries designed to help you use artificial intelligence with confidence.

Be sure to bookmark the Resource Center and check back often as new articles and resources are added.

---

Read more from our sources:

• FBI — Cryptocurrency and AI Scams Bilk Americans of Billions

• FBI Internet Crime Complaint Center — 2025 Internet Crime Report

• Federal Trade Commission — Help the FTC Fight Impersonation Scams

• Federal Trade Commission — Asked to Enter Your Email Address and Password to Open a Party Invite? That’s a Scam

• Federal Trade Commission — New Trends in Reports of Imposter Scams

• Federal Trade Commission — Help Fight Imposter Scams This World Elder Abuse Awareness Day