After over a year of anticipation, OpenAI recently shipped a major update to ChatGPT, adding image-generation capabilities. The AI firm announced the major upgrade during a recent livestream, indicating that users can create detailed images directly from the chat interface.
It’s worth noting that the chatbot’s new image-generation capabilities can create new images or transform existing ones. OpenAI says ChatGPT is now better at text rendering prompt-following and features a smarter use of context.
The feature is rolling out to all Free, Plus, Team, and Pro users in waves and is expected to ship to Enterprise and Education users soon. Moreover, the new image generation capabilities can be accessed from OpenAI’s text-to-video tool, Sora.
While unveiling ChatGPT’s new image-generation capabilities, OpenAI CEO Sam Altman indicated that he had difficulty believing that the images were truly AI-generated.
we are launching a new thing today—images in chatgpt!two things to say about it:1. it’s an incredible technology/product. i remember seeing some of the first images come out of this model and having a hard time they were really made by AI. we think people will love it, and we…March 25, 2025
According to OpenAI’s CEO, Sam Altman:
“This represents a new high-water mark for us in allowing creative freedom. People are going to create some really amazing stuff and some stuff that may offend people; what we’d like to aim for is that the tool doesn’t create offensive stuff unless you want it to, in which case within reason it does. As we talk about in our model spec, we think putting this intellectual freedom and control in the hands of users is the right thing to do, but we will observe how it goes and listen to society. We think respecting the very wide bounds society will eventually choose to set for AI is the right thing to do, and increasingly important as we get closer to AGI. Thanks in advance for the understanding as we work through this.”
ChatGPT’s new image-generation capabilities will be powered by OpenAI’s GPT-4o model. Unlike Dall-e 3, GPT-4o thinks a tad longer to generate more accurate and detailed images.
While speaking to the Wall Street Journal, OpenAI indicated that it trained GPT-4o on publicly available data, including copyrighted data from its partnerships with companies like Shutterstock.
Users are digging ChatGPT’s new image generation capabilities
Since its launch, users have taken to social media platforms to highlight their excitement about ChatGPT’s new image-generation capabilities. As you may know, AI-powered image-generation tools have been swamped with a myriad of issues, from copyright infringement to heavy censorship that seemingly lobotomized their capabilities.
For context, an attempt to get creative with prompts to generate worthwhile outputs is often watered down with an outright rejection due to censorship. Interestingly, reports suggest that AI-powered image-generation tools could be on the precipice of replacing professionals in the built-environment sector, including architects and interior designers.
The report attributed its findings to AI’s insanely great ability to generate sophisticated structural designs. However, a separate report suggests that Copilot and ChatGPT can’t create a simple plain white image.
I attempted to ask ChatGPT to help me generate a simple, plain white image, but its image-generation capabilities are seemingly capped:
“I can’t generate completely blank or plain white images, but you can easily create one using any basic image editor like Paint, Photoshop, or an online tool like Canva. Let me know if you need help with that!”
ChatGPT’s new image-generation upgrade is stunning, and OpenAI has seemingly loosened its grip on its capabilities. A Reddit user recently shared an interesting discovery, indicating that the tool can now generate a glass of wine filled to the brim.
According to a spot by the folks over at Business Insider, ChatGPT’s new image generator tool seemingly had a bug that only allowed it to create images of sexy men, leaving sexy women out of the fold.
According to the software engineer who made the discovery:
“New ChatGPT image gen can draw sexy men but not sexy women. The difference comes down to context and how content is interpreted in terms of sexualization and objectification, especially when it involves women.”
However, the issue doesn’t seem critical, as OpenAI CEO Sam Altman indicated that it was a bug and that a fix would be issued soon.
