OpenAI, the AI startup renowned for its conversational agent ChatGPT, recently announced a significant enhancement to its capabilities by integrating advanced image generation directly into the platform. This update, detailed in a blog post on Tuesday, aims to make image creation a core feature of AI-driven communication. The company’s latest model, GPT-4o, supports this functionality.

In the announcement, OpenAI showcased a sample image on its website, illustrating the improvements made in text accuracy within AI-generated visuals compared to its predecessor models. The initiative seeks to enhance the quality of generated content, particularly by addressing previous limitations related to rendering text and intricate details.

While ChatGPT had previously incorporated image generation through the DALL·E model, this latest update fundamentally improves the user experience, likely offering faster interactions and enhanced features, such as inpainting. OpenAI expressed its commitment to making image generation a primary function of its language models. “At OpenAI, we have long believed image generation should be a primary capability of our language models. That’s why we’ve built our most advanced image generator yet into GPT‑4o. The result—image generation that is not only beautiful but also useful,” read the company’s blog post.

All subscription tiers, including Plus, Pro, Team, and Free, now have access to this new feature.

Sam Altman, the CEO of OpenAI, described the rollout as a leap toward greater creative freedom for users. Acknowledging the potential for controversial outputs, Altman stated, “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.” He further indicated that OpenAI would observe user interactions and feedback closely to navigate this complex landscape.

However, the introduction of the new image generation feature has already sparked controversy. Following the launch, users quickly identified a peculiar glitch. In a social media post, a user claimed that the tool could effectively generate images of “sexy men” but would not create images of “sexy women.” This discrepancy highlights the challenges of moderating AI-generated content, as ethical limits and contextual sensitivities are critical factors.

Altman responded to the situation on social media, acknowledging the issue as an unintended oversight. He confirmed that the team is working to rectify the bug, adding a humorous remark, “Hot guy though!” In a separate testing attempt by Business Insider, it was noted that ChatGPT was capable of generating an image of a “hot woman,” suggesting inconsistencies in the moderation process.

Moreover, OpenAI’s content generation capabilities come with restrictions when it pertains to using images of real individuals. For instance, during a trial run to create an image of Altman himself, the AI could not produce a direct likeness but instead offered a “character drawn from his attributes with a personalized, creative reinterpretation.” Clear guidelines have been established by OpenAI regarding requests involving public figures, especially concerning nudity and graphic content.

Importantly, OpenAI has implemented a policy allowing public personalities to opt out of having their appearances used in AI-generated content. This measure seeks to balance the right to freedom of expression with the principles of privacy and consent as the technology continues to evolve.

Source: Noah Wire Services