OpenAI on Wednesday unveiled Dall-E 3, the latest version of its text-to-image tool that uses its wildly popular AI chatbot ChatGPT to help fill out hints.
Dall-E will be available via API for ChatGPT Plus and enterprise users on October 3, the company said. Users can type a request for an image and correct the prompt by interacting with ChatGPT.
“DALL-E 3 can translate critical applications into highly detailed and accurate images,” the company said in a statement.
OpenAI said the latest version of the tool will have more security measures such as limiting the ability to generate violent, mature or hateful content.
The tool also has mitigations for rejecting requests that ask for images with the name of a public figure, or those that ask for images in the style of a living artist.
Creators can opt out of having some or all of their work used to train future text-to-image tools, OpenAI said.
OpenAI has several competitors in the race to build accurate text-to-image AI tools, including Alibaba’s Tongyi Wanxiang, Midjourney and Stability AI, which continue to improve their image generation models.
However, there are several concerns about AI-generated images. A Washington, D.C. court ruled in August that works of art created by AI cannot be copyrighted under US law.
OpenAI is also facing several lawsuits. A trade group of American writers recently filed a lawsuit against the artificial intelligence leader on behalf of authors including John Grisham and “Game of Thrones” novelist George R.R. Martin, accusing the company of using its chat Botchat is illegally training GPTs on their job.