As a standalone product, Meta is launching Imagine, an AI image generator tool that lets you use text prompts to create images from your desktop or mobile browser. Previously, Imagine was only available in chat apps through the Meta AI virtual assistant.

Imagine uses the Emu technology of Meta’s imaging model. Meta adds invisible watermarks to Imagine to increase transparency and traceability around the AI ​​provenance of the images it creates. The watermark is implemented with a deep learning model and can be detected using the corresponding model. The watermark remains even when you apply image manipulations such as cropping, recoloring, screenshots, and more.

In the future, Meta will bring its invisible watermark to other products that use AI-generated images. At the moment Imagine is intended for use only on US soil.

Source: Android Authority