What you need to know
- Microsoft has announced that Bing AI can now generate images based on natural language prompts.
- Using the Bing Image Creator, powered by DALL-E, users can now simply ask Bing to generate any image they like.
- The new functionality is rolling out in waves, starting today.
Microsoft continues to enhance its new Bing AI Copilot with new features, only a handful of weeks after the service first launched. Today, the company has announced that Bing Chat can now generate images in addition to text, using natural language prompts whenever the users asks for it.
Image creation within Bing is using an "advanced version of the DALL-E model" from OpenAI, and has been available as a standalone service in preview since late last year. Now, users using Bing Chat will be able to simply ask the Bing Copilot to generate images based on their own description and requirements.
Microsoft says there are safeguards in place to stop the generator from creating unsafe images, using OpenAI's own policies and principles as well as its own responsible AI guidelines. Microsoft CVP Yusef Mehdi mentions that the bot will block unsafe images:
When our system detects that a potentially harmful image could be generated by a prompt, it blocks the prompt and warns the user. We also make it clear that Image Creator’s images are generated by AI, and we include a modified Bing icon in the bottom left corner of each image to help indicate that the image was created using Image Creator.
Lastly, the Image Creator feature is coming to Microsoft Edge too, just like Bing Chat. Users will be able to generate images no matter what webpage they're viewing, just by using the Bing sidebar in Edge.
The new Bing has been making waves these last few weeks, with over 100 million users now using Bing to search the web daily. Of that number, over a third of users are interacting with the AI chatbot at least once a day, and Microsoft expects that number to keep growing as the AI becomes more capable.
