OpenAI has postponed the rollout of an “adult mode” for ChatGPT following backlash from its safety advisors. Nevertheless, the developer has not abandoned the concept itself, reports The Wall Street Journal, citing insider sources.
The possibility of introducing this feature had previously been mentioned by CEO Sam Altman, who stated in October that the service might permit adult users to engage in erotic correspondence. However, prior to the launch, the company convened a safety advisory board which unanimously opposed the swift implementation of such a function.
Experts cautioned that this type of AI interaction format could foster an unhealthy emotional dependence among users toward the chatbot. Furthermore, participants in the discussion voiced doubts regarding the reliability of the age verification mechanisms.
According to sources, during testing phases, the age detection system incorrectly categorized minors as adults in approximately 12% of instances. Given the service’s current user base, this could translate to millions of individuals under 18 potentially accessing features intended exclusively for adults.
The developers concurrently struggled to create moderation frameworks that would allow erotic conversations while effectively barring the discussion of prohibited subjects.
Internal deliberations included even more severe criticism. Sources indicate that one board member warned that without stringent safeguards, the product risked becoming a tool with detrimental psychological effects on users.
OpenAI itself insists that the idea is not scrapped, but the feature’s scope will be constrained: the creation of explicit images, audio, or video content is not planned. Moreover, the models will be programmed to periodically remind users about the significance of maintaining human interaction outside the digital realm.
About the Author
