Adobe, a company known for its widely used creative apps such as Photoshop, Illustrator, and Premiere, has launched its own AI image generator called Adobe Firefly. This launch marks a significant move for Adobe as it seeks to remain relevant and competitive in the rapidly evolving creative app ecosystem, where newcomers have been offering powerful tools for creating images, videos, and sound at a lower cost.

The Adobe Firefly is a family of creative generative AI models that enables users to type in a prompt and receive an image in return or generate stylized text. At launch, Adobe is calling Firefly a beta, which will be available through a website (adobe.com/firefly). However, the company plans to integrate the generative AI tools tightly with its suite of creative apps in the future.

One significant twist that Adobe puts on its generative AI tools is that it discloses the data its models are trained on. The company claims that all the data fed to its models are either out of copyright, licensed for training, or in the Adobe Stock library, which gives Adobe’s system the benefit of not infringing on artists’ rights and making its system more brand-safe. The company also plans to compensate artists who contribute training data to its models.

Moreover, Adobe aims to make its AI tools user-friendly and accessible to all, including those without experience in image editing. Adobe includes built-in options for art styles, lighting, and aspect ratio to help users style their images. Additionally, users can apply these effects to an image that’s already generated, eliminating the need to generate a new creation every time an update is needed.

Looking forward, Adobe plans to build these generative tools into its various apps and services. There’ll be AI-generated outpainting in Photoshop; Illustrator will be able to generate vector variations on hand-drawn sketches, while Premiere will allow users to color grade or restyle an image with just an image prompt.

However, Adobe’s plan to let artists train the system on their work to generate content in their style has sparked controversy. Critics argue that the feature could lead to the cloning of artists’ work by others. In response, Adobe is developing a “Do Not Train” system that allows artists to embed the request into an image’s metadata to stop training systems from looking at it.