Meta, the parent company of Facebook and Instagram, has launched a standalone AI image generator called 'Imagine with Meta AI'. The tool, which was trained on 1.1 billion publicly visible Facebook and Instagram images, allows users to generate images from simple text prompts. The AI model can render a novel image from a written prompt, and it can create photorealistic images. However, it struggles with text rendering and different media outputs.
Previously, the tool was only available on Facebook Messenger, but it is now accessible through a standalone website. Meta's AI image offerings also include Emu Edit and Emu Video, which allow users to alter images and generate videos based on text prompts. Additionally, Meta has introduced a feature called Reimagine, which enables users to remix and re-edit AI images created by others in group chats.
To increase transparency and traceability, Meta plans to add watermarks to the generated content. The company is also testing dozens of new generative AI features across Facebook, Instagram, and WhatsApp. These updates highlight Meta's efforts to make generative AI a core part of its service as it competes with other AI companies.
The Meta Imagine service is currently region-locked and may not be available in all countries. Users need to log in with their Meta account to use the service. The generator filters out violence, curse words, sexual topics, and the names of celebrities and historical figures. The pressure is increasing on tech firms to make it clear when works are generated by AI, especially in light of the rise of Deepfakes and AI-generated child abuse images.
