(Reuters) – OpenAI on Wednesday unveiled Dall-E 3, the latest version of a text-to-image tool that uses the wildly popular AI chatGPT to populate prompts.
Dall-E 3 will be available to ChatGPT Plus and Enterprise customers in October via API, the company said. Users can enter an image request and edit the prompt in conversations with ChatGPT.
“DALL-E 3 can transform nuanced requests into highly detailed and accurate images,” the company said in a statement.
OpenAI said the latest version of the tool has more safeguards, such as limiting its ability to create violent, adult or hateful content.
The tool also has ways to reject requests asking for pictures of a public figure by name, or requests asking for pictures in the style of a living artist.
OpenAI said authors could opt out of using some or all of their work to train future text-to-image tools.
OpenAI’s race to create accurate text-to-image AI tools has several contenders, including Alibaba’s Tongyi Wanxiang, Midjourney, and Stability AI, which continue to refine their image creation models.
However, there are several concerns with AI-generated images. A Washington D.C. court ruled in August that artwork created by artificial intelligence without human input cannot be copyrighted under US law.
OpenAI is also facing several lawsuits. A trade group for US authors recently sued the AI executive over the rights of authors including John Grisham and Game of Thrones author George R.R. Martin, who accused the company of illegally training the chatbot ChatGPT in their work.