OpenAI review strategy major adjustments, ChatGPT image generator has caused heated discussion

This week, OpenAI launched a new image generator in ChatGPT, and its ability to generate Ghibli-style images quickly became popular. The generator is based on GPT-4o, which significantly improves ChatGPT's image editing, text rendering and spatial representation capabilities.

However, what is even more striking is the major adjustments to the content review policy by OpenAI. The new policy allows ChatGPT to generate images depicting public figures, hate symbols and racial traits based on user requirements, which is very different from previous practices that strictly restrict controversial content.

Joanne Jang, head of model behavior at OpenAI, said the company is moving from a “full rejection” to a “more precise approach” aimed at preventing real-world harm and stressing “adapting while learning.” OpenAI's move is seen as part of its larger plan to "removal" ChatGPT review, aiming to allow models to process more requests and provide a diverse perspective.

Under the new regulations, ChatGPT can now generate and modify images of public figures such as Trump and Musk. OpenAI believes that it should not be a "arbitrator" who can be generated and provides users with a mechanism to opt out. Furthermore, in educational or neutral settings, ChatGPT can even generate “hate symbols” such as Nazi logos, provided that it is not clear to praise or support extremism. The new model has also relaxed restrictions on requests involving physical characteristics that would have been rejected in the past. At the same time, ChatGPT can also imitate the style of studios such as Pixar or Ghibli, but still restrict imitating living artists.

Nevertheless, OpenAI is not completely liberalizing abuse, and GPT-4o has set stricter safeguards in generating images of children. But the significant changes in content review policies are a time when conservatives have long criticized Silicon Valley companies for their "censorship". Previously, Google's Gemini image generator caused controversy over generating inaccurate images of multiracial historical figures.

Republican Rep. Jim Jordan has questioned technology giants such as OpenAI on the issue of artificial intelligence content review. OpenAI denies that policy changes are politically motivated, saying it reflects “long-term belief in giving users more control” and that technology is mature enough to deal with sensitive topics.

Considering the regulatory scrutiny that may be faced in the future, OpenAI's move is sensitive. Platforms such as Meta and X have also adopted similar strategies to allow more controversial content to be published. Although ChatGPT's new image generator is currently primarily used to create entertaining Ghibli-style images, a shift in its content review policy could have broader implications and could put OpenAI in new challenges in responding to sensitive issues.