AI service loosening restrictions in AI image generation following viral trend reminiscent of Studio Ghibli style: "We're transitioning from strict controls..."
Title: Ghibli-style Art Explosion Unleashes OpenAI's New Image Policies
Author: Vaishali Shastri
Last Updated: March 31, 2025, 09:34 PM IST
Viral Trend echoes OpenAI's loosening grip on content moderation
With the surge of folks experimenting Ghibli-style art using ChatGPT, OpenAI has eased its content moderation guidelines letting the AI model venture into creating images of famous personalities and even controversial symbols, physical features.
In the recent buzz around netizens turning original images into Japanese Ghibli art form, OpenAI has relaxed its content control, allowing its AI model to experiment with pics of celebrities like Elon Musk, Donald Trump, and more. The tech company earlier prohibited such controversial context generation. But partnering with Microsoft, OpenAI now permits ChatGPT to create images for public figures. Joanne Jang, Head of Product and Model Behaviour at OpenAI said, "We're forgoing blanket refusals in sensitive areas for a more specific approach, focusing on averting real-world damage. This move aims at fostering humility, acknowledging our limited knowledge and being open to adapting."
Flaunting Content Limitations
OpenAI, last week, released a white paper, stating users can generate hateful symbols, as long as they're utilized in educational or neutral contexts, not openly promoting extremist agendas. Joanne explained that OpenAI has transformed its definition of offensive content to accommodate changes like altering people's eye color and physical adjustments, deviating from their earlier stance that deemed these alterations as blatantly offensive.
Jang also published a blog post, declaring creation of a customized opt-out list, containing names of public figures who disapprove of their images being utilized by ChatGPT. "AI lab employees should not dictate what people should or should not be allowed to create," the blog highlighted.
A week ago, OpenAI integrated a native image generation tool into ChatGPT through GPT-4o, triggering users to kick off a viral trend, where they prompted AI to transform images into Studio Ghibli style art form.
Trending Now
ghibli viral trendghibli style artstudio ghibli imagesDonald TrumpElon MuskOpenAIopenai chatgpt 4o
OpenAI's policy modifications reflect an evolution in tolerating the creation of images of public figures and controversial symbols, prompted by the advent of the Ghibli-style AI art trend.
Background and Policy Shift
- Evolving Policies: OpenAI's revised policies demonstrate flexibility while balancing user creative freedom with ethical considerations, driven by technical advancements and societal demands[1][2].
- Ghibli Style Art Trend: The policy changes corresponded to a viral trend embracing Ghibli-style AI art, showcasing OpenAI's image generation capabilities[4].
Key Policy Changes
- Public Figures: OpenAI allows the creation of images for public figures like Donald Trump and Elon Musk. A users' opt-out list for such figures has been established[4][5].
- Controversial Symbols: The new policy also allows for the depiction of symbols that were previously restricted, with emphasis on preventing real-world harm[2][4].
- Granular Moderation: OpenAI has shifted from blanket prohibitions to more nuanced content moderation, steering clear of implicit biases in filtering content while focusing on discouraging harmful outcomes[5].
Implications and Challenges
- Creative Freedom vs. Safety: The policy changes reflect a balance between empowering users and maintaining safety measures. Critics worry that expanded freedom may encourage misuse, especially in cases of generating deceptive or harmful content[1][4].
- Regulatory and Social Context: OpenAI's changes coincide with ongoing debates on AI content moderation and potential regulatory pressures. The company emphasizes its dedication to responsible AI deployment[2][4].
- Despite the popularity of the Ghibli-style art trend, OpenAI has now allowed ChatGPT to create images of Bollywood celebrities, signifying an expansion in the scope of figures it can depict.
- The relaxed content moderation policies have enabled OpenAI to allow the creation of images of controversial figures like Donald Trump and Elon Musk, but Joanne Jang has published a blog post for users to opt-out if they wish.
- The viral trend of users turning original images into Ghibli-style art form has prompted OpenAI to allow the generation of images with figures' physical features altered, deviating from its earlier stance.
- With the integration of a native image generation tool into ChatGPT through GPT-4o, India is witnessing a surge in whatsapp conversations discussing the newfound capabilities of the AI model to create a wide variety of images, including those of public figures.
