OpenAI updates content moderation policies on image generation; here’s what allowed now

OpenAI has revised its content moderation policies which enables ChatGPT to generate images of public figures, hateful symbols in educational contexts, and specific racial features upon request. Previously, such requests were declined outright since it was controversial.

OpenAI’s model behaviour lead Joanne Jang explained the policy shift in a blog post. Jang stated that they are shifting from blanket refusals in sensitive areas to a more precise approach which focuses on preventing real-world harm. She added that the goal is to embrace humility, recognising how much they don’t know and positioning themselves to adapt as they learn.

As per a TechCrunch report, this move is a part of OpenAI’s broader strategy to ‘uncensor’ ChatGPT that they announced in February. This strategy aims to reduce refusals and offer diverse perspectives.

OpenAI’s new content moderation policy for ChatGPT

Reportedly, under the new policy, the users can now request modifications in physical features in images like making a person appear heavier or of a different Asian—this was previously prohibited. But OpenAI maintains restrictions on imitating the styles of living artists and still applies the safeguards to prevent misuse, especially regarding children’s pictures.

Also Read : Did Nintendo just drop a big hint about Switch 2? Here's what we know

Access to GPT-4o image generator limited to paid users

According to another Indian Express report, despite its rapid popularity, OpenAI has restricted access to its new GPT-4o image generator to paid users only. Initially, the company promised the feature for free-tier users however overwhelming demand has led to a delay.

CEO Sam Altman, as per the report, in a social media post had written that images in ChatGPT gained more popularity than they anticipated and thus the rollout for their free tier is going to be delayed for a while.

Reportedly, at present, only Plus ($20/month) and Pro ($200/month) subscribers can access the advanced image generator. The users can still generate images, but only using DALL-E. The results of the latter are not on par with those of GPT-4o.

The debate over AI content moderation

As per the TechCrunch report, these changes have come amid growing concern over AI censorship. OpenAI has maintained that these modifications are not politically motivated and that they reflect their long-term goal of giving users more control.

Also Read : Tesco makes changes in app; Clubcard holders must install latest update in 3 days to retain benefits

But OpenAI’s decision could have political and regulatory implications, particularly under the Trump administration which has criticised AI content restrictions. The TechCrunch report added that letting an AI chatbot answer sensitive questions could land the company in hot water soon enough. Tech giants like Meta and X have already relaxed similar policies. This has allowed more controversial discussions on their platforms.

FAQs

1. Why is OpenAI limiting access to GPT-4o's image generator?

Because of excess demand, OpenAI has postponed the release of the GPT-4o image generator for free-tier customers. Currently, it is available to Plus and Pro users only.

2 . What new content can ChatGPT now create?
ChatGPT can now generate images of celebrities, hateful symbols to use for educational purposes, and racial characteristics upon demand. OpenAI still has tight controls in place against objectionable content.