First social media came after teens. Now AI is doing the same thing

Two distinct developments in the tech world are unfolding this week. In a California courtroom, a group of 29 state attorneys general filed a lawsuit against Meta, the parent company of Facebook, Instagram, and WhatsApp. The charges allege that Meta knowingly designed its platforms to addict children, deceived the public about potential harms to young users, and unlawfully collected data from children under 13 without parental consent.

Meanwhile, OpenAI unveiled ChatGPT for Teens, an AI-powered chatbot tailored for users aged 13 to 17, equipped with enhanced safety features, prompts about responsible usage, and additional parental controls. This release comes nearly four years after ChatGPT was launched to the public in November 2022. In contrast, Meta introduced Teen Accounts for Instagram just last September 2024, adding to an array of existing child safety measures.

Both instances illustrate a recurring pattern in the tech industry: companies launch products that entice young users, then establish boundaries around their usage later. OpenAI claims that ChatGPT for Teens incorporates safety measures introduced over the past year, such as parental controls, age prediction, and a version of its Model Spec designed for users under 18.

The company emphasizes prioritizing teen safety and promoting real-world support. However, experts like former Facebook employee Victoria Baines, a professor emerita, observe a similar pattern in both social media and AI safety issues. Baines suggests that companies typically prioritize addressing the most severe risks, such as child abuse and terrorism, which can be detected and prevented at scale.

For risks that aren't necessarily unlawful but can be problematic for users and society, like inappropriate content and misinformation, companies take a more cautious approach. Meetali Jain, executive director of the Tech Justice Law Project, likens generative AI to "social media on steroids," highlighting its potential to deeply influence human interactions.

While social media companies have taken steps to address less obvious safety concerns, the rapid adoption of AI leaves developers with less time to identify and mitigate potential harms compared to social media companies. Experts argue that AI's ability to learn from users' interests and vulnerabilities makes it more insidious. Nina Vasan, director of Brainstorm: The Stanford Lab for Mental Health Innovation, emphasizes that AI companies should face the same scrutiny as Meta regarding their research on young users and the actions they take based on that knowledge.