Open topic · lesson 2 of 3 · 12 min
Deepfakes and what to trust
AI can now produce a convincing voice, face, or video of a real person. Learn how these fakes are used, what the law says, and the habits that still protect you.
- Describe how deepfakes are used to deceive
- Explain why detection alone is not a defense
- Apply verification habits that do not depend on spotting a fake
What changed
A deepfake is audio, an image, or video created or altered by AI to show a real person saying or doing something they did not. What changed is the cost. Cloning a voice once took a studio; now it takes a short recording and an inexpensive tool. In early 2024 an employee of the engineering firm Arup in Hong Kong sent about 25 million dollars to criminals after a video call in which the other participants, including a senior executive, were AI-generated. The same techniques are used for fake family-emergency calls, fake endorsements, and sexual images of real people made without consent.
Audio, an image, or video created or altered with AI so that a real person appears to say or do something they did not.
The FBI advises families to agree on a secret word or phrase that only they know. If a caller who sounds exactly like a relative asks for money in an emergency, ask for the word, or hang up and call the person back on a number you already have. Neither step depends on whether the voice sounds real.
Why you cannot rely on spotting them
Advice to look for odd hands or unnatural blinking is already out of date, and detection software is unreliable against the newest tools. There is a second problem. Once people know convincing fakes exist, a real recording can be dismissed as fake. Researchers call this the liar’s dividend. The practical conclusion is that looking harder is not a defense. Verification has to come from somewhere other than the content itself.
What the law and the industry are doing
In the United States, the Federal Communications Commission ruled in February 2024 that AI-generated voices count as artificial voices under existing robocall law, so such calls require consent. The TAKE IT DOWN Act, signed in May 2025, makes it a federal crime to publish intimate images of a person without consent, including AI-generated ones, and since May 2026 requires covered platforms to remove them within 48 hours of a valid request. Separately, an industry standard called Content Credentials attaches a tamper-evident record of where an image or video came from. It helps when present, and its absence proves nothing.
Set up your verification habits
Put three protections in place for yourself or your household this week.
These habits work against a perfect fake, which is the point. Do not test them by trying to spot fakes.
0 of 5 practice steps checked. Saved on this device.
A family of four chose a code word at dinner and agreed that any request for money by phone gets a call-back, no exceptions. The parents set their social profiles to friends only. They bookmarked the reporting page for the two platforms their teenager uses, and agreed that a surprising clip gets checked against two established news sources before anyone shares it.
Check your understanding
Answer both to mark the lesson complete. You can try again. This is practice, not a graded exam.
1. A caller who sounds exactly like your manager urgently asks you to wire money. What is the most reliable response?
2. What is the liar’s dividend?
Three things to remember
- Convincing fake voices and video are now cheap to make.
- You cannot reliably spot them, so verify through a second channel you already trust.
- Agree on a code word, call back on known numbers, and know how to report a fake.

A plain-language guide to AI for every mind, every age, and every community.
- CNN, Arup revealed as victim of $25 million deepfake scam (May 2024)
- FBI Internet Crime Complaint Center, Criminals use generative AI to facilitate financial fraud (Dec 2024)
- Federal Communications Commission, FCC makes AI-generated voices in robocalls illegal (Feb 2024)
- WilmerHale, The TAKE IT DOWN Act goes live (June 2026)
- Chesney and Citron, Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security (2019)
- Coalition for Content Provenance and Authenticity (C2PA)
Examples are composites written for teaching unless a source is cited. Figures come from the sources listed where given.