← Back to course
Applied AI / Week 06 / Monday
1/6
Week 06 Β· Fakes and Scams

Monday

Scams got a makeover
// Clean text, cloned voices and a family code word
⏱ about 15 min

Monday: Scams Got a Makeover

Maya's neighbor tells her about a strange call. The voice sounded just like her grandson, and he needed help fast.

Ben gets an email that looks like it came from his school office. The logo is right and the writing is perfect.

Rosa gets a message from a "supplier" with clean, friendly wording. It asks her to pay a bill today.

None of these had the bad spelling they were all taught to look for.

What the FBI warns about

In December 2024 the FBI's Internet Crime Complaint Center, called IC3, issued a public alert about generative AI.

It warns that criminals use generative AI to commit fraud on a larger scale. AI makes their schemes more believable.

The alert names fake text, fake images, voice clones and fake video. A voice clone is a fake recording made to sound like a real person.

What a scam can fakeWhat it might look like
TextA friendly, well-written email or text message
ImagesA photo that seems to show a real person or document
VoiceA short audio clip that sounds like someone you love
VideoA clip of a face and voice that seem real

The old warning sign is fading

For years, people were taught to spot scams by their spelling and grammar mistakes.

The FBI alert says AI text can correct the human errors that used to be warning signs of fraud.

So a clean, polished message tells you nothing about who sent it. You need checks that do not depend on how a message looks.

CHECK THE ALERT
  • Read the question.
  • Tap your answer.
Who issued the December 2024 alert about AI and fraud?
What does the alert say AI does to scams?
Why is perfect spelling no longer a sign a message is safe?
StatementTrue or false?
The FBI alert names fake voice clips as one AI scam tool.?
A message with no spelling mistakes is probably real.?
Scams can now use fake images and video, not just text.?
A voice clone is a real recording of the person.?
WHY THIS EXERCISEThe way a message looks is no longer a safe test.

NIST makes a related point about generated content in general. It may not separate fact from opinion or fiction, or say when it is unsure.

A smooth message can sound sure of itself and still be false from start to finish.

In one word: what kind of fake recording sounds like a real person's voice?
WHY THIS EXERCISENaming the trick makes it easier to stay calm when you meet it.
The FBI says criminals use generative AI to make fraud more ____.
AI text can fix the spelling and ____ errors that used to warn us.
Generated content may not say when it is ____.

Sources: FBI IC3, Criminals Use Generative AI to Facilitate Financial Fraud (2024) Β· NIST AI 600-1, Generative AI Profile (2024)

Tomorrow: the voice on the phone, and four habits that still work.