Image Generation

Today's Stories

1. 1

   AI food ads spark backlash; restaurants say they're temporary

   Restaurants are using generative AI to create food ads instead of photographing real meals. Examples in New York City and San Francisco have drawn criticism online, with one X post gaining 2.8 million views and a Reddit post about a "trypophobia burrito" earning over 670 upvotes. Professional-looking food ads no longer require professional photography, which helps small businesses save time and money. However, AI can depict meals that never existed, raising questions about when an ad becomes deceptive—a concern experts like Harvard Law professor Rebecca Tushnet and advertising lawyer Libby O'Neill say depends on whether the image makes specific claims about the product.

   The test is whether customers accept imperfect real photos over polished fake ones, and whether regulators or courts clarify the line for AI-generated food imagery. The 1965 Supreme Court case FTC v. Colgate-Palmolive Co. shows the issue isn't new, but AI's ability to create photorealistic fakes may push the boundary further.

2. 2

   Why AI food images look so unappetizing

   Restaurants, cafes, and brands are using AI to generate food images, but the results often look like 'slop'—for example, donut shrimp, wormlike noodles, and ice cream that resembles concrete. Diffusion models (which generate images by gradually removing noise) first recover coarse structures, then add fine details—this can corrupt the basic shape early on. Also, these models 'reproduce looks without proper knowledge about the world,' so they have no idea what a sandwich or burrito actually is.

   The test is whether AI can overcome its weakness in generating the details that make food recognizable — thin, continuous, terminating structures that currently produce "spaghetti-like artifacts" and visual degeneration.

3. 3

   AI-generated restaurant menus look off, and science explains why

   AI-generated menus in restaurants often feature food that looks too perfect, symmetrical, and smooth, triggering unease in customers. This stems from AI models trained on a narrow, 'pleasing' aesthetic. Researchers at the University of Duisburg-Essen found these images create an 'uncanny valley' effect, where almost-real food looks more disgusting than obviously fake images. The homogenization happens because models optimize for 'pleasingness' and 'not being offensive,' which 'shaves off the edges.'

   The test is whether restaurants will keep using AI menus after the uncanny valley effect triggers customer backlash, or whether the trust issue spreads beyond food. Watch whether Alex Lisle’s Reality Defender and similar AI-detection startups gain traction as visual evidence credibility erodes.

4. 4

   Why AI images feel "creepy": a PR expert explains

   An ITmedia article explains why consumers find AI-generated images "creepy" (a common Japanese slang term for unsettling), with a public relations professional providing the analysis. For businesses and communicators, understanding this consumer discomfort matters because it affects how AI imagery is perceived in marketing and public communications, potentially influencing brand reception.

   The article is presented as the first of two pages, so the explanation continues in the second part.

5. 5

   Fei-Fei Li's World Labs unveils Atlas 3D world model

   World Labs, founded by Stanford professor Fei-Fei Li, unveiled Atlas, a multimodal world model that can generate, reconstruct, and simulate 3D environments from text, images, and video. This moves AI beyond 2D content into full 3D simulation, which could be important for robotics and immersive applications. The model is designed to help machines understand and interact with physical spaces, not just generate media.

   The test is whether Atlas moves from research demos to real-world adoption, where robotics and embodied-AI developers stand to gain the most. Watch for the first commercial or enterprise integrations announced after the HumanX Conference 2026.

6. 6

   Jasper Research releases text-to-image build cookbook

   Jasper Research released a cookbook explaining how to build a text-to-image model from scratch, sharing full reasoning and intermediate results. The cookbook is positioned as ideal for deep-diving into text-to-image models or for learning how frontier labs build them, making an advanced process accessible to a broader audience.

   The cookbook’s value hinges on whether readers can actually reproduce the training process with the included 100M-image dataset and tiny-model codebase, not just follow the reasoning. Watch for community attempts to run the nano t2i codebase in the coming weeks.