During a livestream on Tuesday, Openai CEO Sam Altman announced the first major upgrade to Chatgpt's Image generation abilities for more than one year.
Chatgpt can now use the company GPT-4O Model to native create and change images and images. GPT-4O has long supported the AI-powered chatbot platform, but to this day, the model has been able to produce and only edit text-not images.
Altman said the GPT-4o native generation of image is now live in Chatgpt and Sora, Openai video-generation product, for subscribers to the company's $ 200-a-month pro plan. Openai said the feature was rolling as soon as possible to plus and free chatgpt users, as well as developers with the company's API service.
GPT-4O with image output “thinks” longer than the image generation model that it effectively replaces, From 3To do what Openai described as more accurate and detailed images. GPT-4O can edit existing images, including images with people with them-they change these or “inpainting” details such as facade and background objects.
In order to empower the new image feature, Openai said in Wall Street Journal It trained the GPT-4O on “available public data,” as well as ownership data from its cooperation with companies such as shutterstock.
Many generative AI vendors see training data as a competitive advantage, so they keep it and any information related to it near the chest. But training data details are also a potential source of IP-related suits, another unpleasant for companies to reveal many.
“We respect the rights of artists in terms of how we do the output, and we have the rules in the area that prevent us from developing images that directly imitate any living artist,” said Brad Lightcap, Openai's chief operating officer, in a journal statement.
Openai offers an opt-out form that allows creators to request to remove their works from training datasets. The company also says it respects requests that web-scraping bots will not allow the collection of training data, including images, from websites.
The upgraded feature of ChatGPT image-image follows the heel of Google's native image experimental image for Gemini 2.0 Flash, one of the company's flagship models. The strong feature has gone viral on social media – but not necessary for the best reasons. Part of the Gemini 2.0 Flash image has been Some guardsallowing people to remove watermarks and create images that describe characters with copyright.
This article was updated at 12:00 pt to include Openai's statement in the Wall Street Journal around GPT-4O training data.
