What AI cant fake about the real world (yet) @ScienceMagazine
What AI cant fake about the real world (yet)  @ScienceMagazine
Uploaded June 2026 | Updated November 2026, 1 day ago
Manipulating a photo used to take skill, but now a convincing fake is just a prompt away. Science reporter Kai Kupferschmidt visited Hany Farid, a pioneer of digital forensics, to find out how he’s keeping his edge in the era of generative AI. From shadows that don't line up to sound that doesn't match a scene, Farid breaks down his everyday work of teasing apart what’s real and what’s fake. They discuss where they think the future is heading—and how it's not too late to choose another path.

0:00 Introducing Hany Farid
0:56 Analyzing footage from a missile strike in Iran
3:30 How AI gets the physics wrong (and why we don't notice)
7:00 The pitfalls of AI "enhancement"
8:32 What will the future look like?

Read the full profile: science.org/content/article/deepfakes-are-everywhere-godfather-digital-forensics-fighting-back
What AI cant fake about the real world (yet)Origin and radiation of squids revealed by digital fossil-miningThese beetles wear chemical ‘invisibility cloaks’—and will die if they’re removedThis robot hugs objects into placeA vision for the future needs scientistsMoonlight drives energy balance for these little birdsHow scientists measure Atlantic currentsHow Jeff Horwitz gained access to Metas researchLotions and fragrances hamper the body’s protective oxidation cloudThis rock-eating ‘worm’ could change the course of riversNew flu preventative could outperform vaccinesDoes this wolf want to play—or attack? Take a close look at its face
Science Magazine |

What AI can't fake about the real world (yet)

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER