In response to Google’s Gemini launch, Meta introduces Imagine with Meta, a standalone generative AI experience available on the web.
Imagine with Meta allows users to create images by describing them in natural language, similar to OpenAI’s DALL-E. This standalone experience, akin to Google’s Gemini, generates high-resolution images based on text prompts, leveraging Meta’s Emu image generation model.
Despite the creative potential of such tools, Meta has faced criticism for past issues, including a racially biased AI sticker generator. The company aims to address concerns by implementing safeguards in Imagine with Meta. While the tool wasn’t available for testing before its launch, Meta acknowledges the need for careful monitoring as Imagine with Meta reaches a broader user base.
To enhance transparency, Meta plans to introduce watermarks to content generated by Imagine with Meta. These invisible watermarks, developed with an AI model, will be detectable using a corresponding model. Although the detection model’s public release is not confirmed, Meta assures resilience to common image manipulations, such as cropping, resizing, color changes, and more.
The move towards watermarking for generative art aligns with industry trends. Other companies, like Imatag and Steg.AI, have developed watermarking tools that are resilient to various image manipulations. Meta’s commitment to implementing invisible watermarking extends to its future products featuring AI-generated images.
Amid increasing pressure on tech firms to address the clarity of AI-generated content, Meta’s decision to add watermarks is a step towards meeting regulatory expectations. Recent regulations in China and discussions in U.S. Senate committee hearings underscore the importance of transparency in generative AI, with watermarks emerging as a practical solution to distinguish works created by AI.
As the technology landscape evolves, Meta’s Imagine with Meta aims to balance creativity with responsible AI practices.