← L² Lab
🧠 Critical Thinking
Card 30
🤖 📹 ❓

How can you tell if a video, image, or text is AI-generated?

💭 How to Think About This

That video of a politician saying something shocking? Deepfake. That amazing photo? AI-generated. That convincing article? Written by ChatGPT. In 2025, AI can create realistic fake content. How do we know what's real anymore? Let's develop detection skills!

🔒 Start writing to unlock hints

DEEPFAKES = AI-generated or manipulated media (video, audio, images, text) that looks/sounds real but isn't! Face-swapping, voice cloning, generated images, fake articles. Technology is INCREDIBLY good now - but not perfect! Detection requires careful observation!

Unnatural movements: Weird blinking, stiff expressions

Lighting inconsistencies: Shadows don't match

Edge artifacts: Blurry boundaries around face/body

Too perfect: Skin too smooth, teeth too white

Hand problems: Extra fingers, weird proportions (AI struggles with hands!)

Background glitches: Warped objects, impossible geometry

Audio deepfakes: Robotic cadence, breathing sounds missing, emotional tone doesn't match words, background noise inconsistent.

AI text: Too perfect grammar, generic/vague details, repetitive phrasing, lacks personal voice, no verifiable sources, sounds like it's summarizing rather than experiencing!

1. Reverse image search: Google Lens, TinEye

2. Check source: Is this from official channel?

3. Corroboration: Do other credible sources report this?

4. Context check: When/where was this supposedly recorded?

5. Metadata analysis: File properties, creation date

6. Ask experts: Fact-checking sites (Snopes, FactCheck.org)

If something seems too shocking/perfect - VERIFY before sharing!

AI-generated and manipulated content is everywhere in 2025 - develop critical detection skills!

Types of AI content:

Deepfake videos: Face/body swapping, lip-syncing

Voice clones: AI mimics anyone's voice

Generated images: Completely fabricated photos

AI text: Articles, posts, comments by bots

Hybrid: Real footage + AI manipulation

Detection checklist for VIDEO:

✓ Watch facial movements - natural blinking?

✓ Check lighting consistency

✓ Look at edges and boundaries

✓ Observe hands and fingers carefully

✓ Check background for warping

✓ Listen for audio sync issues

Detection checklist for IMAGES:

✓ Zoom in on details (AI makes small errors)

✓ Check hands, teeth, jewelry

✓ Look for repeated patterns (AI copies)

✓ Examine shadows and reflections

✓ Reverse image search

Detection checklist for TEXT:

✓ Too perfect? Suspiciously polished?

✓ Generic without specific details?

✓ Check for verifiable facts

✓ Does it have authentic voice/personality?

Golden rule:

If something seems too outrageous, too perfect, or too convenient - PAUSE. Verify before believing. Definitely verify before SHARING!

Ask: Who benefits from this being fake? Who benefits from this being real?

🤔 Which thinking lens(es) did you use?

Select all the lenses you used:

👨‍👩‍👧

Adult Guidance

Story Seed: "Look at this crazy video of [celebrity] saying [shocking thing]!" "Hold on - let me check. See how their face edges look weird? And their hands have 6 fingers? This is a deepfake. Let's verify before we share."
Discussion Guide
  • Hands game: Look at AI images together - count fingers, check proportions
  • Reverse image search: Practice using Google Lens on suspicious images
  • Source verification: Where did this actually come from originally?
  • The "too perfect" test: Real life is messy - too polished = suspicious