Image Generation Using One or More Neural Networks

USPTO has published patent application US20260099963A1 titled "Image Generation Using One or More Neural Networks" by inventor Ming-Yu Liu. The application, published April 9, 2026 with a filing date of December 11, 2025, covers techniques for training neural networks to generate and manipulate digital images, including modifications based on user-selected features. The invention relates to the field of neural network computing (CPC G06N) and image processing (CPC G06T, G06V).

For technology companies, researchers, and manufacturers developing AI-based image generation or manipulation systems, this published application represents prior art that may affect freedom-to-operate analyses. Patent applications do not yet confer enforceable rights but establish priority dates for any resulting patents. Parties engaged in neural network development or image synthesis technologies should monitor the prosecution of this application to assess potential infringement risks upon grant.