YouTube, the video platform owned by Alphabet’s Google, began testing a new artificial-intelligence “age-estimation model” in the United States on Wednesday. The system infers whether a logged-in user is under 18 by analyzing viewing history and other account signals, ignoring the birth date supplied at sign-up. The pilot initially reaches only a small slice of U.S. accounts, mirroring technology the company says already works in several other countries. Accounts that the model judges to belong to minors will automatically shift to existing safeguards: access to age-restricted videos is blocked, personalized advertising is disabled, and digital-wellbeing features such as screen-time and bedtime reminders are enabled. Users who believe they have been misclassified can verify their age by uploading a government-issued ID, a credit card or a selfie, according to a company blog post by James Beser, YouTube’s director of product management for youth products. Civil-liberties advocates including the Electronic Frontier Foundation and the Center for Democracy & Technology say the program could compromise privacy and chill speech, while creators warn that a larger pool of teen-categorized viewers may cut into ad revenue. The trial arrives amid mounting political and legal pressure on tech platforms to better shield children online, following the U.S. Supreme Court’s June decision upholding a Texas law requiring age verification for adult websites.