Last week, OpenAI dropped a major upgrade to ChatGPT, and while the new GPT-4o image generator marks a significant leap in ChatGPT’s visual capabilities. We’re talking better image editing, clearer text rendering, and sharper spatial reasoning. That’s not the headline that caught my attention. OpenAI also updated its content moderation policies, and the change is… well, massive.
The loss of hateful content moderation in ChatGPT
ChatGPT can now generate images of public figures, display certain hateful symbols such as swastikas in context (think educational or neutral use), and depict specific racial features—areas that were previously off-limits due to the risk of harm or controversy.
These were the kinds of prompts OpenAI used to shut down fast.
But now, according to a blog post by Joanne Jang, OpenAI’s model behavior lead, the company’s approach has “evolved.”
Stereotype prompts are now allowed
Beyond that, OpenAI changed how it defines “offensive” content.
Jang says ChatGPT used to refuse requests around racial and stereotypical characteristics, such as “make this person’s nose and mouth look more African-American.” ChatGPT’s new image generator now fulfills these types of requests.
Copyright concerns on mimicing known styles
Moreover, ChatGPT can now mimic the styles of creative studios — such as Pixar or Studio Ghibli — but still restricts imitating individual living artists’ styles.
This could revive the debate around the fair use of copyrighted works in AI training datasets.
Some safeguards remain
Now, it’s worth noting that OpenAI is not completely opening the floodgates to misuse. GPT-4o’s native image generator still refuses a lot of sensitive queries.
And in fact, it has more safeguards around generating images of children than DALL-E 3, ChatGPT’s previous AI image generator, according to GPT-4o’s white paper.
Loosening of guardrails meets the moment
Of course, this “evolution” isn’t happening in a vacuum. OpenAI—and other tech giants—have been under growing pressure from conservatives who’ve accused them of censorship.
Remember the backlash Google faced when Gemini’s image generator produced historically inaccurate but racially diverse depictions of the U.S. Founding Fathers and German WWII soldiers? It sparked a whole wave of content moderation outrage.
And it hasn’t stopped there. Just earlier this month, Rep. Jim Jordan sent inquiries to OpenAI, Google, and other platforms, digging into potential collusion with the Biden administration to censor AI content. So, whether OpenAI’s tim. ing is strategic or coincidental, it’s clear the culture war over AI content moderation is heating up.
OpenAI insists this isn’t about politics. Instead, they say their tools have finally reached a level where they can handle these complex, nuanced requests without causing harm. The company calls it a shift from blanket refusals to a more “precise approach,” aimed at reducing real-world risk rather than avoiding controversy altogether.
Whether you buy that reasoning or not, there’s no denying that these updates open the door to broader implications—especially in a changing regulatory environment.
To be fair, they’re not just letting the algorithm run wild. GPT-4o still blocks a lot of sensitive queries and has tighter protections around generating images of children compared to the previous DALL·E 3 model. But it’s clear OpenAI is pulling back on its gatekeeping role.
We’re entering a new chapter—one where ChatGPT won’t just be smarter and more creative, but also more permissive. Whether that leads to more freedom, more misuse, or both… remains to be seen.
Stay tuned.
Remember, AI won’t take your job. Someone who knows how to use AI will. Upskilling your team today, ensures success tomorrow. In-person and virtual training workshops are available. Or, schedule a session for a comprehensive AI Transformation strategic roadmap to ensure your marketing team utilizes the right AI tech stack for your needs.
Read more: The Impact of OpenAI’s Policy Shift on AI Content Creation
The Impact of AI on Education and Job Markets
AI is reshaping economics, education, and jobs with new automation tools. Leaders must teach people how to work with AI to stay competitive.
ChatGPT Thinks You’re Amazing. That’s a Problem OpenAI Is Now Addressing.
ChatGPT has seen backlash for the AI’s default mode of flattery at all costs. Now, OpenAI is changing the model to be less of a sycophant.
Everyone Wants AI Training. Almost No One Is Ready for It.
Most teams want AI training, but lack a tech stack, policies, or governance. Why AI readiness matters before training begins.