The model represents a leap forward in the ability to generate images that exactly adhere to the text provided by the user. It is built natively on ChatGPT, which lets users use ChatGPT as a brainstorming partner and refiner of their prompts. When prompted with an idea, ChatGPT will automatically generate tailored, detailed prompts for DALL·E 3 that bring the idea to life. If a particular image is not quite right, users can ask ChatGPT to make tweaks with just a few words.
Like previous versions, OpenAI has taken steps to limit DALL·E 3’s ability to generate violent, adult, or hateful content. It has mitigations to decline requests that ask for a public figure by name. OpenAI has also improved safety performance in risk areas like generation of public figures and harmful biases related to visual over/under-representation. They are also researching the best ways to help people identify when an image was created with AI and are experimenting with a provenance classifier—a new internal tool that can help identify whether or not an image was generated by DALL·E 3. DALL·E 3 is designed to decline requests that ask for an image in the style of a living artist. Creators can now also opt their images out from training of future image generation models.