Skip to content

Meta Introduces Standalone AI-Powered Image Generator

Meta's new tool, 'Imagine with Meta,' lets users create images via text. With AI safeguards and invisible watermarks, it ensures transparency in AI-generated content.

Meta Launches 'Imagine with Meta': Text-to-Image Creation Tool

In response to Google's Gemini, Meta steps into the limelight with 'Imagine with Meta,' an independent generative AI experience available on the web. This tool, similar to OpenAI's DALL-E and powered by Meta's Emu image generation model, allows users to craft high-resolution images through natural language descriptions.

Originally embedded within Meta's messaging platform for playful interactions, 'Imagine with Meta' extends its reach beyond chats, enabling free image creation on the web. Despite Meta's past issues with image generation tools, the company assures safeguards for this new release.

To enhance transparency and traceability, Meta plans to introduce invisible watermarks to content generated by 'Imagine with Meta' in the upcoming weeks. These watermarks, generated using AI models, aim to resist common image alterations and manipulations, ensuring content authenticity.

Meta emphasizes the resilience of these watermarks to various image modifications, including cropping, resizing, color adjustments, and even sticker overlays. The company envisions expanding this watermarking technique across its AI-generated content in the future.

The tech industry faces mounting pressure for clear labeling of AI-generated content, prompted by concerns about Deepfakes and AI-generated abusive imagery. Both global and regulatory initiatives urge tech companies to adopt measures ensuring transparency in AI-generated creations, leading to the incorporation of watermarking technologies and standards by firms like Meta, Microsoft, Google, and others.

Recent regulations from China's Cyberspace Administration and discussions in U.S. Senate committee hearings, notably by Senator Kyrsten Sinema, underscore the need for transparency and traceability in generative AI, emphasizing the adoption of watermarks as a means of identification without impacting user experience.