Types of fake and where they appear

Telling truth from AI fakes · Lesson 2 / 20

Different fakes, different checks

There's no universal technique. The method depends on the type of fake, and the type is identified by what someone wants from you.

Synthetic text

Fake news, fabricated reviews, "personal stories", mass comments manufacturing the appearance of opinion. It appears wherever volume matters: marketplaces, social networks, comment sections.

Check: look for verifiable facts and the primary source. Style will tell you nothing.

Generated images

A "photo of an event" that never happened; non-existent people for fake profiles; fabricated screenshots of conversations and documents.

Check: reverse image search, and look for independent shots of the same event from other angles.

Video deepfakes

A well-known person saying something they never said. Most often used to advertise dubious "investments" and to discredit people.

Check: look for the same statement in the person's or organisation's official channels. If a dramatic statement exists only in one clip, it wasn't made.

Voice fakes

A call from a "relative" in trouble, "the bank", "your manager" asking for an urgent transfer. The most dangerous type, because it targets emotion and leaves no time.

Check: call back yourself, on the number you already have. Always.

Combined schemes

The most dangerous form is the combination: an email that knows details about you, plus a call in a familiar voice, plus a convincing website. Each element alone would raise doubt; together they corroborate one another. That's precisely the design.

Insight. The elements of a scheme corroborate each other: the email makes the call plausible, the call makes the website convincing. Each channel has to be checked separately and independently.

Common mistake. Assuming fakes only concern politics and celebrities. The bulk of the volume is reviews, fake profiles and targeted calls to ordinary people.

Pro tip. Identify the type of fake before checking it. Primary-source search works for text and a call-back works for voice; confuse them and you'll waste the effort.

Cheat sheet

Text — find the primary source; style is useless.
Images — reverse search and independent angles.
Video — look in official channels.
Voice — call back yourself, always.

1. How do you check a dramatic statement in a video?

2. What makes a combined scheme dangerous?

3. What makes up the bulk of fakes?

Task — checked by AI

Find, in your own feeds, inbox or messengers from the past week, one example of at least three of the fake types in this lesson (synthetic text, generated image, video, voice, or a combined scheme — whatever you encounter). For each, identify the type and name the specific method you'd use to check it. Check one and describe the result.

← Back

🔒 Answer the question correctly to move on to the next lesson.