In response to Google’s Gemini launch, Meta is launching Imagine with Meta, a new stand-alone generative AI online experience that lets users create images by describing them in natural language.
Imagine with Meta, which is fueled by Meta’s current Emu image generating model, generates high-resolution pictures from text prompts in a manner akin to OpenAI’s DALL-E, Midjourney, and Stable Diffusion. Users in the United States can use it for free (for the time being) and it produces four photos for each question.
We’ve liked learning from users how they use Meta AI’s text-to-image generating tool, imagine, to create amusing and imaginative conversation material. We are extending the ability to envision outside chat rooms today,” Meta says in a blog entry that was posted this morning. “Our messaging system is meant for more lighthearted, back-and-forth exchanges, but you can also make free images online.”
Now that Meta’s image production capabilities have recently gotten the firm into trouble (see: Meta’s AI sticker generator that is racially biassed), this writer is left wondering if Imagine with Meta has any protections in place to make sure that history doesn’t repeat itself. Although we weren’t able to test the tool before it launched, you can be sure that we’ll be closely monitoring Imagine with Meta as it gains traction.
Though they won’t go live right now, Meta promised that in the upcoming weeks, they would start watermarking Imagine with Meta material for “increased transparency and traceability.” (A watermark is already evident.) According to Meta, the invisible watermarks will be created using an AI model and identifiable using a matching model. There’s been no news on when the detection model will be released to the public.
“Commonly used image manipulation techniques like cropping, resizing, colour modification (brightness, contrast, etc.), screen grabs, image compression, noise, sticker overlays, and more do not affect the watermarks,” Meta wrote in the article. “In the future, we hope to add invisible watermarking to a large number of our products that use AI-generated images.”
Techniques for watermarking generative art are not new. Imatag, a French firm, says that their watermarking tool is unaffected by picture resizing, cropping, editing, or compression. Watermarks that withstand resizing and other modifications are applied using an AI model by Steg.AI, another company. While Shutterstock and Midjourney have agreed to rules to contain marks indicating their material was developed by a generative AI tool, Microsoft and Google have embraced standards and technology for watermarking that are based on artificial intelligence.
However, there is increasing demand on tech companies to disclose that their creations were created using artificial intelligence (AI). This is especially true in light of the proliferation of Deepfakes from the Gaza War and the filter-bypassing of AI-generated photographs of child abuse.
China’s Cyberspace Administration has released laws mandating that providers of generative AI label information created by the technology, such as text and picture generators, without compromising user experience. Additionally, Senator Kyrsten Sinema (D-AZ) underlined the necessity of openness in generative AI at recent U.S. Senate committee hearings, including the use of watermarks.