Meta strengthens child safety efforts with new AI tools to detect exploitation content and hidden abuse networks
Meta says it removed or took action against 33.2 million pieces of child sexual exploitation content across Facebook and Instagram during the first half of 2026, as the company expands its use of artificial intelligence to identify harmful activity faster.
According to Meta, more than 97% of the content it acted on was detected by its own systems before users reported it. In India alone, the company took action against 5.3 million pieces of child sexual exploitation content in the same period, with over 98% found proactively.
The company is now introducing new AI-based tools designed to identify suspicious ads and accounts that may be used to quietly direct people to illegal child abuse material outside Meta’s platforms. Meta says bad actors are increasingly using ads that appear harmless on the surface but lead users to harmful destinations elsewhere on the internet.
To address this, Meta has developed a new large language model system to detect what it calls “signposting.” This refers to content, especially ads, that may not contain illegal material directly but may guide users toward websites or services that host or share child sexual abuse material.
Meta says it is no longer focusing only on the contents of an ad. The company is also examining where the ad sends people after they click. If a destination violates Meta’s rules, the company can block the website, stop the ad from running, and take action against the accounts involved.
The latest system is part of a broader push to improve child safety detection across Facebook, Instagram, WhatsApp, and Meta’s AI products. Meta says it is adding more AI-driven scans to find harmful content that older detection systems may have missed. The company also plans to keep updating these systems as it learns more about how abuse networks operate and attempt to avoid enforcement.
Another tool Meta is introducing is a red-teaming AI agent. This system is designed to test Meta’s own safety protections by looking for weaknesses that criminals could exploit. By simulating possible abuse tactics, Meta hopes to identify vulnerabilities before they are used more widely.
The company is also improving its ability to detect people who try to return to its platforms after being removed. This includes users who create new accounts after previous accounts were banned for violating child safety rules.
Meta’s latest announcement comes as the company continues to face intense scrutiny over child safety on social media. Regulators, lawmakers, and families have raised concerns about the risks young users may face online, including exposure to harmful content, exploitation, and unsafe interactions.
Earlier this year, Meta agreed to pay up to $18 billion to settle a child safety lawsuit involving 29 U.S. states. The case focused on concerns about how children are protected across the company’s platforms.
Meta has also introduced several child safety features throughout the year. These include parental controls for Meta AI, preteen accounts on WhatsApp, and alerts for parents when children search Instagram for self-harm related content.
WhatsApp has also received new parental control options. Parents can limit how teenagers use Channels, manage who can view their children’s status updates, and control who can add them to groups. They can also choose to receive notifications about certain group activity.
With these updates, Meta is positioning artificial intelligence as a major part of its child protection strategy. The company says faster detection, stronger account enforcement, and closer monitoring of suspicious links and destinations are key to stopping exploitation networks before they can spread harmful content.
