Meta announced on Wednesday that it took action against 33.2 million pieces of child exploitation content on Facebook and Instagram in the first half of 2026. The company also rolled out new AI tools to detect ads and accounts that covertly direct users to illegal child exploitation material. More than 97% of the content actioned was found by the company’s systems before any user complaint was filed. In India, action was taken against 5.3 million pieces of content during the same period, and more than 98% of them were likewise detected proactively.
Meta’s new large language model (LLM) system targets a tactic it calls “signposting”: ads that appear harmless but are suspected of directing users to illegal content off the platform. Even though the ads themselves do not contain illegal material, they can serve as a gateway to sites hosting child exploitation content. For this reason, Meta now looks not only at the content of an ad but also at where it leads the user; it can block sites that violate its rules and take action against the accounts behind them.
The company’s other new tools:
- Additional AI scans that find content previous systems may have missed
- A “red team AI agent” that tests weaknesses in Meta’s own safety measures
- Improved systems that detect those who return to the platform with new accounts after their accounts have been shut down
Meta stated that it will continue adding new signals as it learns how these networks operate.
The new measures come at a time when Meta is under pressure from lawsuits and criticism over child safety. In August, the company agreed to pay up to $18 billion in a lawsuit brought by 29 US states. Safety features introduced this year included:
- Parental controls for Meta AI
- Preteen accounts on WhatsApp
- Alerts to parents when children search for self-harm content on Instagram
In September, WhatsApp gave parents the ability to limit teens’ use of Channels, determine who can see status updates, decide who can add their children to groups, and receive notifications about certain group activities.
Background
Meta is not a new name in the FikirPilot archive: we have published 39 stories mentioning this name in the last 90 days; the most recent is dated October 8, 2026.