Social Media Platforms Expand AI Age Checks

Social media platforms are increasingly estimating users’ age groups from behavior and account information, then applying protections for younger users. In April 2025, Meta began testing an AI system in the United States to identify Instagram accounts likely belonging to teenagers even when registered with an adult birthday, and switch them to Teen Accounts. In September 2025, it expanded the test to the United Kingdom, Canada and Australia. Japan has offered Teen Accounts since January 2025, but an AI-driven automatic switch has not been confirmed there. Reports say the estimates use factors such as the kinds of content users interact with, profile information and when an account was created. Meta says it continues to improve accuracy and provides a way for people wrongly classified to change their settings.

Teen Accounts are for users ages 13 to 17. They automatically limit messages to people a user follows or is already connected with, and enable features such as Sleep Mode, which silences notifications from 10 p.m. to 7 a.m. Users under 16 need parental approval to weaken the protections. Content standards being phased in across Japan from May 2026 draw on film ratings for audiences ages 13 and older. Posts containing strong language or dangerous stunts may be hidden even when shared by someone the user follows; accounts that repeatedly post content deemed unsuitable for their age cannot follow one another or exchange direct messages. Parents can also choose the stricter “Limit Content” setting.

Unlike Instagram, Discord first applies teen settings to everyone and then determines who is an adult. It began introducing this global default in March 2026. Sensitive images are blurred, and only users verified as adults can unblur them or enter age-restricted servers and channels. Messages from strangers are sorted into a separate inbox. Discord’s model considers account age, device, usage and patterns across the community; when it cannot determine an age, it asks for an estimate from a face photo or for identity documents. Roblox has required face-based age estimation before chat use worldwide, including in Japan, since January 2026. ChatGPT also began a global rollout that month of a feature that predicts from usage whether someone is under 18 and automatically applies protections for minors.

Age estimation can sometimes avoid collecting identity documents, reducing the risk of leaks involving names, addresses, face photos and ID numbers. In September 2026, unauthorized access to Times Car obtained information from about 6.6 million accounts, including former members; images of driver’s licenses, student IDs and other documents were also leaked for about 1.6 million accounts. Discord said in October 2025 that about 70,000 identity-document images submitted for age checks may have been exposed through a contractor.

Age estimation can also misclassify people. Australia’s government-commissioned Age Assurance Technology Trial concluded that the technology was broadly accurate and practical, while finding that errors were more likely around threshold ages. In a test with 16 as the cutoff, 8.5% of people who were actually 16 were classified as under 16, according to reports. Many services require a face photo or identity document to correct a mistaken classification, leaving users with the risks of submitting official documents. A newly 18-year-old high school senior may use social media much like a 17-year-old student, making it difficult to distinguish them accurately from behavior data alone. Some have therefore called for age checks when people sign mobile phone contracts or use app stores.

When children register as adults, protections such as limits on messages from strangers, hiding intense content and parental supervision do not work. Since August 2026, Instagram has offered a notification to parents in Japan who use parental controls if their child repeatedly searches within a short period for terms related to serious distress. Accurate age registration is a prerequisite for the feature. The article says children who falsify their age may hesitate to tell parents about problems because they fear their rule-breaking will be discovered. It also cites cases of children who could no longer livestream on Instagram without parental permission registering with a false age on TikTok, where users must be at least 18 to livestream. In Japan, an expert panel at the Children and Families Agency is expected to complete its final report by the end of the year, as policymakers work on rules including mandatory age checks for social media companies.

Social media platforms are increasingly estimating users’ age groups from behavior and account information, then applying protections for younger users. In April 2025, Meta began testing an AI system in the United States to identify Instagram accounts likely belonging to teenagers even when registered with an adult birthday, and switch them to Teen Accounts. In September 2025, it expanded the test to the United Kingdom, Canada and Australia. Japan has offered Teen Accounts since January 2025, but an AI-driven automatic switch has not been confirmed there. Reports say the estimates use factors such as the kinds of content users interact with, profile information and when an account was created. Meta says it continues to improve accuracy and provides a way for people wrongly classified to change their settings.

Teen Accounts are for users ages 13 to 17. They automatically limit messages to people a user follows or is already connected with, and enable features such as Sleep Mode, which silences notifications from 10 p.m. to 7 a.m. Users under 16 need parental approval to weaken the protections. Content standards being phased in across Japan from May 2026 draw on film ratings for audiences ages 13 and older. Posts containing strong language or dangerous stunts may be hidden even when shared by someone the user follows; accounts that repeatedly post content deemed unsuitable for their age cannot follow one another or exchange direct messages. Parents can also choose the stricter “Limit Content” setting.

Unlike Instagram, Discord first applies teen settings to everyone and then determines who is an adult. It began introducing this global default in March 2026. Sensitive images are blurred, and only users verified as adults can unblur them or enter age-restricted servers and channels. Messages from strangers are sorted into a separate inbox. Discord’s model considers account age, device, usage and patterns across the community; when it cannot determine an age, it asks for an estimate from a face photo or for identity documents. Roblox has required face-based age estimation before chat use worldwide, including in Japan, since January 2026. ChatGPT also began a global rollout that month of a feature that predicts from usage whether someone is under 18 and automatically applies protections for minors.

Age estimation can sometimes avoid collecting identity documents, reducing the risk of leaks involving names, addresses, face photos and ID numbers. In September 2026, unauthorized access to Times Car obtained information from about 6.6 million accounts, including former members; images of driver’s licenses, student IDs and other documents were also leaked for about 1.6 million accounts. Discord said in October 2025 that about 70,000 identity-document images submitted for age checks may have been exposed through a contractor.

Age estimation can also misclassify people. Australia’s government-commissioned Age Assurance Technology Trial concluded that the technology was broadly accurate and practical, while finding that errors were more likely around threshold ages. In a test with 16 as the cutoff, 8.5% of people who were actually 16 were classified as under 16, according to reports. Many services require a face photo or identity document to correct a mistaken classification, leaving users with the risks of submitting official documents. A newly 18-year-old high school senior may use social media much like a 17-year-old student, making it difficult to distinguish them accurately from behavior data alone. Some have therefore called for age checks when people sign mobile phone contracts or use app stores.

When children register as adults, protections such as limits on messages from strangers, hiding intense content and parental supervision do not work. Since August 2026, Instagram has offered a notification to parents in Japan who use parental controls if their child repeatedly searches within a short period for terms related to serious distress. Accurate age registration is a prerequisite for the feature. The article says children who falsify their age may hesitate to tell parents about problems because they fear their rule-breaking will be discovered. It also cites cases of children who could no longer livestream on Instagram without parental permission registering with a false age on TikTok, where users must be at least 18 to livestream. In Japan, an expert panel at the Children and Families Agency is expected to complete its final report by the end of the year, as policymakers work on rules including mandatory age checks for social media companies.