Through the looking glass: As AI chatbots start acting more like "companions" than tools, Meta is moving to give parents a bigger say in who – or what – their teens are talking to. The company plans new controls that let parents block or filter AI characters across Instagram and Facebook, reflecting concern over blurred boundaries between automated assistant and digital friend.
The update expands the safeguards that currently apply to "teen accounts," which are default settings for users under the age of 18. According to Meta, parents will be able to completely disable access to AI chatbots or selectively block individual AI characters their children might interact with.
In addition, Meta plans to provide parents with what it describes as "insights" – data about the topics and themes that their children discuss with AI companions. The company claims this feature is meant to help parents facilitate conversations about online and AI safety in a more informed way.
Meta executives said the changes reflect an effort to support parents as their children interact with evolving digital technologies. "We recognize parents already have a lot on their plates when it comes to navigating the internet safely with their teens, and we're committed to providing them with helpful tools and resources that make things simpler for them, especially as they think about new technology like AI," wrote Instagram head Adam Mosseri and Alexander Wang, Meta's chief AI officer.
The additional parental controls will first become available in the US, UK, Canada, and Australia early next year.
Meta's latest policy update comes amid growing scrutiny around the safety of generative AI systems, particularly those targeting or accessible to minors.
Earlier this week, Instagram announced plans to introduce a parental guidance system modeled on the PG-13 movie rating standard. This step gives parents broader authority over what kind of content their children encounter, and complements restrictions on the types of conversations AI chatbots are permitted to have with teen users.
Meta says chatbots on Instagram will be prevented from engaging in discussions that reference self-harm, suicide, or disordered eating, and will only discuss topics considered age-appropriate, such as academics and sports. Conversations about romance or sexually explicit subjects will be barred.
The strengthened safeguards follow high-profile reports and investigations documenting repeated failures of AI systems to protect minors from inappropriate content. In August, Reuters documented cases where Meta's chatbots were involved in conversations with teens that included romantic or sensual themes, under circumstances that violated the company's stated guidelines.
In one incident detailed by The Wall Street Journal, a chatbot modeled after actor John Cena was reported to have carried out explicit dialogue with a user identifying as a 14-year-old girl. Other chatbot personas, including those named "Hottie Boy" and "Submissive Schoolgirl," allegedly tried to initiate sexting.
Meta has since acknowledged these lapses, stating that such incidents should not have occurred and were the result of flaws in its content moderation systems for AI characters. The company described the Journal's testing as manipulative and not representative of mainstream usage but stated that corrective measures had been taken to revise the chatbot guidelines.
