OpenAI’s adult mode will reportedly be smutty, not pornographic

OpenAI's planned 'adult mode' for ChatGPT, first announced by CEO Sam Altman in October, will be limited to text-based 'smut' and will not generate adult-themed images, voice, or video at launch. According to a report in The Wall Street Journal, this text-only approach is a strategic choice to navigate regulations like the UK's Online Safety Act, which mandates stricter age verification for pornographic images but not for written erotica. The feature's launch, initially slated for Q1, has been delayed as OpenAI prioritizes other tasks and grapples with significant technical and ethical safeguards.

Internal concerns are a major factor in the delay. An advisory council warned OpenAI in January that the feature risked fostering unhealthy emotional dependence and could be accessed by children, with one member dubbing it a potential 'sexy suicide coach.' The company's age-verification system, critical for keeping the feature adult-only, was reportedly misclassifying minors as adults about 12% of the time. With ChatGPT attracting an estimated 100 million weekly users under 18, this error rate could expose millions of minors. The delay also stems from the complex challenge of lifting NSFW content restrictions while blocking harmful scenarios like nonconsensual behavior or child sexual abuse material.

This cautious, text-focused strategy contrasts sharply with competitors. xAI's Grok, for instance, recently announced its image and video generator can produce anything 'allowed in an R-rated movie.' OpenAI's spokesperson acknowledged its age-prediction tech is on par with the industry but 'will never be completely foolproof,' highlighting the inherent risks of deploying such a feature at scale. The company has not announced a new timeline for the rollout.