[iTokio]

The trick is not so much about compression, but rather about image generation. The trade off is completely different from usual lossy algorithms. A highly compressed image might still retain high visual quality but with completely different details, textures.

Kinda what would happen if you use a perfect painter with a blurry memory.

[Tossrock]

This is quite visible in their demonstration slider-image; the building in the back-right changes from a brick building with glass windows to a stucco-ish building with a bunch of exterior duct-work.

[fludlight]

Also, the car behind the bus disappears. It looks like it's been photoshopped out. This is unexpected behavior from a compression algorithm. Users are conditioned to expect the quality to degrade uniformly across the whole image.

[Veedrac]

In fairness that detail isn't preserved in the other formats either. The new issue is merely the illusion of accuracy.

[rasz]

car is visible in WebP.

[w_t_payne]

You would have to be very very careful indeed if you wanted to use this for anything with safety implications.

[frenchie4111]

This would have interesting implications if we accidentally trained biased networks (provocative example: improperly changing people's skin-colors in images when decoding them)