LESSON 5 · Don't Get Fooled: Media Literacy
Detection Clues That Still Work
Deepfakes keep improving, but current versions may still show unnatural blinking, waxy skin, or lighting that does not match. One clue proves little; several together matter.

For AI-generated text, watch for perfect grammar with no personality — it can be technically flawless yet emotionally flat. Also look for repetitive sentence structures, since models tend to fall into the same patterns. The clearest tell is often confidence without substance: claims stated firmly but with no specific, checkable detail behind them.