ChatGPT Image Generation Upgraded

# ChatGPT # openai

In a livestream on Tuesday, OpenAI's CEO, Sam Altman, revealed a significant enhancement to ChatGPT's image-generation capabilities, marking the first major update in over a year. Now, ChatGPT can harness the power of the GPT-4o model to create and modify images and photos directly. Although GPT-4o has been the backbone of the AI chatbot platform for some time, its abilities were previously limited to text generation and editing. Altman announced that the native image generation feature powered by GPT-4o is now available for subscribers of OpenAI's $200-a-month Pro plan, both in ChatGPT and Sora, the company's AI video-generation tool. OpenAI plans to extend this feature to Plus and free users of ChatGPT soon, as well as to developers using its API service. Compared to the DALL-E 3 model it replaces, GPT-4o takes a bit more time to generate images but produces results that are more accurate and detailed. Additionally, GPT-4o can edit existing images, including those with people, allowing for transformations and detailed "inpainting" of foreground and background elements. To develop this new image feature, OpenAI utilized "publicly available data" along with proprietary data from partnerships with companies like Shutterstock, as reported to the Wall Street Journal. The specifics of training data are often kept under wraps by AI companies due to competitive advantages and the risk of IP-related lawsuits. Brad Lightcap, OpenAI's chief operating officer, emphasized the company's commitment to respecting artists' rights. "We’re respectful of the artists’ rights in terms of how we do the output, and we have policies in place that prevent us from generating images that directly mimic any living artists’ work," he stated to the Journal. OpenAI also provides an opt-out form for creators to request the removal of their works from its training datasets and respects requests to prevent its web-scraping bots from collecting data, including images, from websites. This upgrade to ChatGPT's image-generation capabilities comes shortly after Google introduced experimental native image output for its Gemini 2.0 Flash model. However, Google's feature quickly drew attention on social media, not always positively, as it lacked sufficient guardrails, allowing users to remove watermarks and generate images of copyrighted characters. *This article was updated at 12pm PT to include OpenAI’s statement to the Wall Street Journal regarding GPT-4o’s training data.*

Related article

Comments (52)