Amid viral trend of turning images into Ghibli style art form, OpenAI has eased its content moderation policies which has now enabled its AI model to experiment with pictures of famous personalities, and even discriminatory symbols and physical features.
Amid Ghibli style viral trend, Sam Altman’s OpenAI relaxes rules of AI image generation: ‘We’re shifting from…’
As netizens have been experimenting with ChatGPT’s image generation tech by creating Ghibli style art and turning original images into the Japanese art form, OpenAI has eased its content moderation policies which has now enabled its AI model to experiment with pictures of famous personalities, and even discriminatory symbols and physical features.
The tech firm had earlier disallowed to generate images of public figures in such controversial context. But the company, with partnership with Microsoft, has allowed ChatGPT to create images of public personalities such as Elon Musk and Donald Trump, and others. Joanne Jang, Open AI’s head of product and model behaviour said, “We’re shifting from blanket refusals in sensitive areas to a more precise approach focused on preventing real-world harm. The goal is to embrace humility: recognizing how much we don’t know, and positioning ourselves to adapt as we learn.”
Users can now generate images with controversial symbols
OpenAI also released a white paper last week, in which it made clear that users can “generate hateful symbols,” but in educational and neutral context which don’t “clearly praise or endorse extremist agendas.” According to Jang, Open AI has evolved its meaning of offensive content, including in it prompts such as changing people’s eye colour, and those around physical changes, a stance that has differed from its earlier position where the company found such changes as outright offensive.
She also published a blogpost in which he mentioned that a customised list, an opt-out list, has been created which has names of public figures who do not permit their images to be used by ChatGPT. “AI lab employees should not be the arbiters of what people should and shouldn’t be allowed to create,” the blog read.
A week back, OpenAI built a native image generation tool in ChatGPT through GPT-4o, which encouraged users to start a viral trend which saw them prompting AI to turn images into Studio Ghibli style art form.
