OpenAI launched DALL-E 3 with improved context understanding and ChatGPT integration

Just a month and a half after its development was revealed, OpenAI has launched DALL-E 3 in research preview, the latest version of its AI visual art platform, featuring enhanced contextual understanding to fix issues with specific text prompts from previous versions. This iteration incorporates ChatGPT Plus, enabling users to create prompts with less need for detailed instructions. With this integration, ChatGPT can generate more extended paragraphs as prompts, enhancing the image quality produced by DALL-E 3.

The goal of this integration is to make AI art creation accessible to a broader user base. Dall-E was first introduced in January 2021, paving the way for similar text-to-image AI platforms. DALL-E 3, launched in 2022, had a waitlist due to concerns about generating explicit and biased content. DALL-E 3 will initially be available to ChatGPT Plus and ChatGPT Enterprise users in October, with future releases planned for research labs and its API service in the fall.

For DALL-E 3, OpenAI has implemented strong safety measures, including mechanisms to avoid the generation of inappropriate or offensive images. By working with external red teamers and using input classifiers, the model is trained to disregard certain words, thus preventing explicit or violent prompts. DALL-E 3 will also not reproduce images of public figures if their names are used in the prompts. To address copyright issues, OpenAI allows artists to exclude their art from future text-to-image AI models, responding to copyright infringement lawsuits filed by artists against competitors and art websites.