ChatGPT’s new image generator goes viral amid privacy and ethical concerns
OpenAI's latest upgrade to ChatGPT's image generation capabilities has taken the internet by storm, but the viral success is accompanied by growing concerns regarding copyright, ethical boundaries, and content moderation policies.
Unveiled during a livestream by CEO Sam Altman last week, the new model, GPT-4o, allows users to natively create and modify images directly within ChatGPT’s interface. Altman stated that the feature is initially available for subscribers to the $200-a-month ChatGPT Pro plan, with plans to extend access to Plus and free users, as well as developers using the company's API.
GPU overload
Within hours of the launch, social media platforms were flooded with AI-generated images, particularly in the iconic style of Studio Ghibli, the beloved Japanese animation studio behind classics like My Neighbor Totoro and Spirited Away.
The response was so overwhelming that Altman took to X (formerly Twitter) to announce that OpenAI’s GPUs were "melting" due to the demand.
"It’s super fun seeing people love images in ChatGPT. But our GPUs are melting. We are going to temporarily introduce some rate limits while we work on making it more efficient," Altman wrote, adding that free-tier users would soon be limited to three generations per day.
Copyright and legal concerns
While the viral success of ChatGPT’s image generator has sparked excitement, it has also reignited debates over AI-generated art and copyright infringement. Many experts question whether AI tools trained on copyrighted works violate intellectual property laws.
Evan Brown, an intellectual property lawyer at Neal & McDevitt, notes that while artistic style itself is not explicitly protected by copyright, concerns arise over how AI models are trained. If OpenAI used copyrighted Studio Ghibli films to train its model, it could potentially face legal challenges, similar to the ongoing lawsuits against OpenAI, Google, Meta, and Midjourney for allegedly using copyrighted materials without permission.
In response, an OpenAI spokesperson stated that ChatGPT does not allow users to replicate “the style of individual living artists, but it does permit imitation of 'broader studio styles.” However, this stance remains controversial, as many living artists are credited with pioneering the unique styles of their respective studios.
In the case of Studio Ghibli, it’s Hayao Miyazaki who explicitly expressed his distaste for AI-generated content a few years ago.
Changes in OpenAI’s content moderation policies
Beyond its new creative capabilities, OpenAI has also adjusted its content moderation policies. ChatGPT will now allow users to generate images of public figures, certain hateful symbols in educational contexts, and racial features—an area the company previously restricted.
“We’re shifting from blanket refusals in sensitive areas to a more precise approach focused on preventing real-world harm,” OpenAI’s model behavior lead, Joanne Jang, wrote in a blog post.
Under the updated policy, users can now generate images of figures like Donald Trump and Elon Musk, which were previously restricted. OpenAI argues it does not want to act as an "arbiter of status" and is offering an opt-out for individuals who do not want to be depicted.
While OpenAI has relaxed some guardrails, it emphasizes that GPT-4o's image generator still has safeguards in place, including stricter measures around generating images of children compared to its previous model, DALL-E 3.
The future of AI
The viral success of ChatGPT's new image generator has undoubtedly captured the public's imagination, but it also offers a glimpse into a scary future where AI piggybacks on hard-acquired human skills and talents at the expense of specific individuals. That’s on top of misuse of the technology for malicious activities like enhancing conspiracies and propaganda.
