Chatgpt’s generation of image mode receives upgrade

During a lifetime on Tuesday, Openai’s CEO Sam Altman announced the first major upgrade to Chatgpt’s image capabilities in more than a year.

Chatgpt can now take advantage of the company’s GPT-4O model to create and modify the company’s images and photos. The GPT-4O has long supported the AI ​​Chatbot platform, but so far the model has managed to create and edit only the text-not the images.

Altman said the GPT-4O inherent image production is alive today in Chatgpt and Sora, Openai’s Video Generation AI product for subscribers to Pro Pro’s $ 200-A-Month. Openai reports that the feature is soon released on Plus and free Chatgpt users, as well as developers using the company’s API service.

The GPT-4O output of the image “thinks” slightly more than the image-producing image model that effectively replaces, Dall-E 3, to do what Openai describes as more accurate and detailed images. The GPT-4O can edit existing images, including images with people in them-transcending them or “inpainting” details such as foreground and background objects.

To feed the new image mode, Openai told Journal Wall Street She trained GPT-4O on “available in common data” as well as privately owned data from her partnerships with companies such as Shutterstock.

Many AI genetic suppliers see training data as a competitive advantage, so that they can maintain any information related to the chest. However, IP training data is also a possible source of IP pipelines, another disincentive to companies to reveal a lot.

“We respect the rights of artists in the way we make the production and have implemented policies that prevent us from creating images that directly mimic every work of living artists,” said Brad Lightcap, head of Openai, in a statement to the magazine.

Openai offers an exemption form that allows creators to ask for their projects to be removed from the training data sets. The company also states that it respects the requests to prohibit tissue posting bots from the collection of training data, including images, from websites.

The upgraded function of Chatgpt’s upgraded image follows the heels of Google’s Experimental Image Exit for Gemini 2.0 Flash, one of the company’s flagship models. The powerful trait became viral in social media – but not necessarily for the best reasons. The element of the Gemini 2.0 Flash image has proven to have few protective messages, allowing people to remove watercolors and create images depicting copyright -protected characters.

This article was updated at 12pm PT to include Openai’s statement in the Wall Street Journal around GPT-4O training data.