Images of global figures like President Donald Trump, Chinese President Xi Jinping, and Tesla CEO Elon Musk can now be freely generated and modified using ChatGPT.
This shift comes as OpenAI introduces its ChatGPT-4o Image Generation feature, which has sparked a Ghibli-style trend while significantly relaxing restrictions on AI-generated content.
IT industry sources reported that Joanne Zhang, OpenAI’s Head of Model Behavior, shared on her X account on March 27 that the ChatGPT-4o image-generation feature marks a special release. She highlighted CEO Sam Altman’s description of it as “a new high-water mark for us in allowing creative freedom.”
Zhang emphasized the company’s trust in user creativity over internal predictions, stating that OpenAI employees should not act as gatekeepers, deciding what users can or cannot create. Following the tool’s release, she humbly acknowledged that users often discover applications beyond initial limitations.
Zhang explained that the goal is to maximize creative freedom while preventing actual harm. OpenAI aims to apply policies fairly and equally, regardless of an individual’s status. Instead of determining who qualifies as “important enough” to be protected, the company has opted for an “opt-out list,” allowing individuals to request restrictions on their image being generated.
This approach enables the AI model to mimic anyone’s image, including public figures, while respecting personal preferences for exclusion.
Addressing concerns over “offensive content,” Zhang acknowledged that OpenAI previously lacked clear guidelines, which led to rejecting prompts such as “make this person’s eyes look more Asian.” She noted that such decisions were unintended consequences of restrictive policies and justified the company’s move toward a more lenient stance.
Zhang admitted that while OpenAI may be perceived as lowering safety standards, this policy shift does not undermine the team’s extensive research and ethical considerations. Instead, the company is adopting a more permissive approach to foster innovation.
Foreign media reports indicate that tests have confirmed the AI now accepts requests to create Nazi symbols, modify facial features to appear more Asian, or adjust body size, raising ethical concerns.
TechCrunch warned that AI content moderation could become a battleground for cultural and political disputes.
This move follows OpenAI’s decision in February to relax ChatGPT content restrictions by gradually removing the orange warning boxes.