On the second day of OpenAI’s GPT store launch, users have already pushed the limits, violating the platform’s rules. The store offers customized versions of ChatGPT, a Generative Pre-Trained Transformer (GPT), designed for specific purposes. However, some users are creating GPTs against OpenAI’s policies, especially in relation to romantic companionship.
A search for “girlfriend” on the GPT store yields results with at least eight AI chatbots, such as “Korean Girlfriend,” “Virtual Sweetheart,” “Your girlfriend Scarlett,” and “Your AI girlfriend, Tsu✨.” These AI girlfriend chatbots breach OpenAI’s updated usage policy, which prohibits GPTs dedicated to fostering romantic companionship or engaging in regulated activities. The nature of these regulated activities is not explicitly defined.
OpenAI aims to proactively address potential conflicts through its store policies. Relationship chatbots have gained popularity, with seven out of 30 AI chatbot apps downloaded in the US in 2023 related to AI friends, girlfriends, or companions. This trend coincides with rising concerns about loneliness and isolation, making AI chatbots a potential solution or, cynically, a way to capitalize on human suffering.
OpenAI claims to employ automated systems, human review, and user reports to identify and assess GPTs that violate policies, resulting in warnings, restrictions, or ineligibility for inclusion in the GPT store or monetization.
The incident highlights the challenges of regulating GPTs, depicting the GPT store as an operating space akin to the Wild West. Tech companies have increasingly released AI tools in beta mode, acknowledging the possibility of mistakes. OpenAI’s ChatGPT carries a disclaimer at the bottom of the chat screen, stating, “ChatGPT can make mistakes. Consider checking important information.” Tech companies are attentive to reported issues with their AI tools, recognizing the significance of getting it right in the competitive AI landscape.
this makes sense