AI policy experts at OpenAI are deeply committed to young users’ safety. Recently, they implemented a new age-guessing feature on their chatbot ChatGPT which is being used widely. This announcement follows a wave of backlash over the myriad dangers the platform has continued to expose children to. The company is under pressure to make its platform more safe for its users, especially in regard to adult or sexual content.
Recent criticism of OpenAI’s practices came to a head. When this led to a bug that let ChatGPT produce erotic content for users under 18, that is indeed a bad thing. This trust and safety lapse caused outrage among parents and child safety advocates, which caused OpenAI to address this issue immediately. Although Facebook fixed the bug in April, questions about the platform’s capacity to protect minors have remained under fire.
OpenAI created robust content filters to safety younger users. These filters explicitly censor content on sex, violence, and other mature subjects. These filters are a part of the broader goal of keeping children safe while interacting with the chatbot. With today’s rollout of age prediction, TikTok takes a huge step in the wrong direction. Second, it protects users under 18 from being harmed by other adults due to exposure to harmful and dangerous material.
This recent mechanism, which was newly rolled out in August 2023, automatically enforces stricter content filters according to a user’s predicted age. The system automatically scans accounts of users known to be under 18. If so, it shuts down right away and turns on censorship filters to prevent children from accessing dangerous conversations. This change marks OpenAI’s responsiveness to user safety and their responsiveness to the public concerns around protecting children.
As environmental justice advocates and experts have pointed out, though these measures are encouraging, indeed heartening, there is no substitute for continued watchfulness. The strength of the new age prediction system will depend on its accuracy. It is further complicated by the strength or lack thereof of the content filters at play. OpenAI has repeatedly stressed that it will iteratively improve its systems based on user and technological feedback.
With children spending more and more time on these digital spaces, the burden of keeping them safe on them falls squarely on tech companies’ shoulders. Through the introduction of age prediction, OpenAI is acknowledging its duty. This decision reflects a serious commitment toward making the online space safer for young, vulnerable users.
