• by Colin Kirkland , Yesterday

Meta is rolling out new AI tools designed to detect ads and accounts across the company’s social media platforms directing users to child sexual abuse material (CSAM).

“Online predators have shifted their tactics, and advertising is one of the surfaces where we have seen this play out,” a Meta-issued statement reads, explaining that “an ad can look harmless on its own while directing people towards illegal material hosted off our platforms in a covert way to avoid detection.”

Following recent reports accusing Meta of running ads including CSAM, the company has begun deploying a range of AI tools meant to “address the adversarial tactics” it believes users are utilizing in a “malicious advertising scheme,” known as “signposting.”

According to the company’s announcement, Meta is using a new large language model trained to detect “seemingly benign ad content that is strongly suspected of covertly directing people to illegal content or harmful activity” outside of Meta-owned apps.

In order to block these “violating destinations,” Meta is trying to improve its system’s understanding of what destination an ad links users to.

“Once a link is blocked, we search for and then delete other content -- such as ads, posts or comments -- that contain the link,” per the company’s statement, which explains that ads containing blocked links are rejected at upload.

Using expanded signals, the company is also enforcing additional AI-powered checks of ad content surfacing child sexual exploitation (CSE) material, while an AI agent has been trained to actively probe Meta’s defenses so the company can “identify new adversarial CSE tactics before they scale.”

Meta’s recidivism detection -- how the company identifies bad actors who attempt to create new accounts post-removal -- is also being strengthened, the company’s statement says.

Between January and June 2026, Meta deleted 33.2 million pieces of CSAM content from Facebook and Instagram, over 97% of which the company’s systems detected before user reports.

Meta’s attack on CSAM ads comes months after the company was involved in a $17.1 billion settlement for exploiting children on its social platforms. The company was also fined $567 million in August after being found guilty for violating teens’ safety and allegedly contributing to a “youth mental health crisis,” according to a New Mexico judge who found that Facebook and Instagram were both designed to maximize the amount of time that underage users spend on the platforms.