OpenAI CEO Sam Altman has announced a significant update to GPT-4o, the advanced AI model that powers ChatGPT, marking a notable advancement in AI capabilities. This announcement was made on March 28, 2025, via X (formerly Twitter), where Altman described the update as a "big stride" forward in AI performance, particularly in its ability to tackle complex technical and coding problems.
In his tweet, Altman emphasized that the new version of GPT-4o is "particularly good at coding, instruction following, and freedom." This enhancement is designed to make GPT-4o more efficient at handling intricate coding challenges, improving code quality, and executing detailed instructions with greater accuracy. Users can expect the model to respond better to multi-part instructions and to produce fewer emoji-based responses.
OpenAI echoed Altman’s sentiments in a tweet, stating, "GPT-4o got another update in ChatGPT! What's different? Better at following detailed instructions, especially prompts containing multiple requests. Improved capability to tackle complex technical and coding problems. Improved intuition and creativity. Fewer emojis.” This latest update is currently available to paid ChatGPT users, with free users expected to gain access soon.
Moreover, developers can utilize the improved model through OpenAI’s API, allowing for seamless integration into various applications. The excitement surrounding this update reflects the growing demand for advanced AI functionalities, as users increasingly rely on AI for coding and technical tasks.
In addition to the update on GPT-4o, OpenAI has also announced changes to its content moderation policies concerning image generation. As of March 29, 2025, OpenAI is shifting from blanket refusals in sensitive areas to a more nuanced approach aimed at preventing real-world harm. Joanne Jang, OpenAI’s model behavior lead, explained this policy shift in a blog post, emphasizing the company's commitment to embracing humility and adapting as they learn.
This policy change is part of OpenAI’s broader strategy to 'uncensor' ChatGPT, which was first announced in February 2025. Under the new guidelines, users can now request modifications in physical features in generated images, such as altering a person to appear heavier or of a different Asian ethnicity. However, OpenAI has maintained restrictions on imitating the styles of living artists and continues to apply safeguards to prevent misuse, particularly regarding images of children.
Despite the rapid popularity of the new GPT-4o image generator, access has been limited to paid users only. Initially, OpenAI promised free access to this feature, but overwhelming demand has led to a delay in its rollout for free-tier users. Currently, only Plus subscribers ($20/month) and Pro subscribers ($200/month) can access the advanced image generator, while other users can still generate images using DALL-E, albeit with results that do not match those produced by GPT-4o.
OpenAI’s decision to update its content moderation policies comes amid growing concerns over AI censorship. The company has asserted that these modifications are not politically motivated and reflect their long-term goal of providing users with more control over the content they create. However, the implications of this decision could be significant, particularly in light of regulatory scrutiny and political commentary surrounding AI technologies.
As tech giants like Meta and X relax their content policies, the pressure is on OpenAI to balance innovation with responsibility. Allowing an AI chatbot to answer sensitive questions could lead to potential controversies, and as Jang noted, the company is conscious of the need to navigate this landscape carefully.
In summary, the recent updates to GPT-4o and OpenAI's content moderation policies represent a pivotal moment in the evolution of AI. With enhanced capabilities in coding and image generation, OpenAI is positioning itself at the forefront of AI technology while also grappling with the ethical implications of its advancements. As users await broader access to these features, the ongoing dialogue about AI's role in society and the responsibilities of its creators remains more relevant than ever.
