Meta Uses AI to Detect Ads Linked to Child Sexual Exploitation

Jakarta (ANTARA) – Meta is developing artificial intelligence (AI)-based features to detect accounts and advertisements suspected of directing users to material involving child sexual exploitation outside its platforms.

According to a TechCrunch report on Wednesday (7 October), local time, Meta acted against 33.2 million items involving child sexual exploitation on Facebook and Instagram during the first half of 2026. More than 97 per cent of the content was detected by Meta’s systems before users reported it.

The security system under development uses a large language model (LLM) to detect a practice the company calls “signposting”. This refers to advertisements that appear normal but are actually used to direct users to illegal content or harmful activities outside Meta’s platforms.

Information about the advertisements’ destinations can help Meta block websites that breach its rules and take action against accounts involved in signposting.

Meta has also added AI-based scanning to find child exploitation content that may have been missed by its detection systems. It has updated its systems to identify people who create accounts on its platforms after a previous account was removed for violating its rules.

These measures come as Meta continues to face criticism over child safety on its social media platforms.

In September, WhatsApp also added parental controls for teen accounts, including restrictions on using Channels, settings for who can view status updates, and controls over who can add children to groups.