Just as we predicted, OpenAI is ready to launch DALL-E 3, the latest version of its generative AI visual art platform. And to make things interesting, this version will be integrated with ChatGPT, which will enhance prompt creation by understanding the context better and accurately translating text prompts into images. 

Integration with ChatGPT will allow users to create prompts swiftly without needing to craft detailed instructions. The aim is to simplify the creative process and increase user engagement with AI art.

The phased release of DALL-E 3 will start with ChatGPT Plus and ChatGPT Enterprise users in October, followed by research labs and the API service in the fall. OpenAI remains committed to further enhancing safety measures and iterating on the model, aligning with advancements and evolutions in AI technologies.

In a demo, lead researcher Aditya Ramesh showcased the collaboration between ChatGPT and DALL-E 3, illustrating the seamless generation of logos for a ramen restaurant set in mountainous landscapes. This demonstration emphasised how this integration enhances creative potential, especially for individuals with varying artistic abilities.

OpenAI has diligently refined the safety aspects of DALL-E 3. The system now implements measures to prevent the generation of lewd, offensive, or potentially harmful images. Additionally, restrictions have been placed on recreating images resembling public figures, maintaining ethical standards and privacy concerns. DALL-E 3 has been engineered to refuse requests seeking an image resembling the style of a currently active artist. Additionally, creators now have the option to exclude their images from being used in training their upcoming image generation models.

Recently, OpenAI made a significant announcement, inviting domain experts to join the OpenAI Red Teaming Network — an initiative to improve the safety of their models. The announcement seeks to ‘deepen collaborations’ with external experts and research institutions.