Microsoft has integrated OpenAI’s latest DALL-E 3 text-to-picture model into Bing Image Creator and Bing Chat, reports The Register.
“The DALL-E 3 model from OpenAI delivers enhancements that improve the overall quality and detail of images, along with greater accuracy for human hands, faces, and text in images,” the company explained.
DALL-E 3 better understands the prompts and generates images that more accurately reflect the user’s preferences. At the same time, the model will not be able to create harmful or inappropriate images that reflect violence or “toxic” content.
Meanwhile, the created images will now have a watermark with the date and time. There will also be information that the work was generated by artificial intelligence.
The DALL-E 3 understands Ukrainian language well and produces fairly accurate images based on a free description in Ukrainian.
Earlier, experts warned about the risks of using AI tools to create fake images. Therefore, an appropriate mark may be the key to solving the problem. However, some experts still doubt the effectiveness of such watermarks.
Microsoft is also aware that text-to-picture technology poses some challenges. But it assures that its team is working on solving them.
“As always, we encourage customers to practice good habits online, including exercising caution when providing sensitive personal information,” the company urged.
In the summer, Bing Chat started to compete with Google Bard in Google Chrome and Apple Safari.