Overview
- The viral images that spread in August and early September moved the trend from online posts into printed posters and storefront menus, producing widespread public disgust and documented harm to vendors.
- Psychologists say near‑real but flawed AI food pictures hit an uncanny valley effect and can provoke strong aversive reactions, including trypophobic responses that make people avoid a menu item.
- Technical experts trace the problem to diffusion-based image models that struggle with thin, terminating structures like noodles and repeated textures, and to training-data biases that amplify strange artifacts.
- Restaurants face real costs from the backlash: vendors report negative reviews, social media shaming, removal or cleanup expenses, and 26% of owners already use AI tools, making the risk widespread.
- Industry advice is to use real smartphone or professional photos or to apply AI only to enhance real images while researchers warn that feeding model outputs back into training sets can cause visual homogenization and quality decline.