OpenAI has announced a major update to image generation capabilities in ChatGPT. The new features allow the GPT-4o model not only to create text, but also to generate and edit images. This update is available to ChatGPT Pro plan users, and will soon become available to Plus and free subscribers, as well as developers using the API.
The GPT-4o model, which replaced DALL-E 3, delivers more accurate and detailed images. It can edit existing images, including those featuring people, by changing or adding details. OpenAI reported that the model was trained using publicly available data and data obtained from partner companies such as Shutterstock.
The company emphasizes respect for artists’ rights and has introduced a policy that prevents the creation of images that directly imitate the work of living artists. OpenAI also allows creators to remove their works from training datasets and block data collection by their web scraping bots.
This update follows experimental attempts by other companies, such as Google, in the field of image generation. OpenAI highlights the implementation of robust safeguards against misuse, including watermark removal and the creation of inappropriate images. All generated images contain standard C2PA metadata marking them as created by OpenAI, and users have the right to use them in accordance with the company’s policy.
