Meta deploys AI tools to detect child abuse content after 5.3 million violations in India

Meta AI tools detect child abuse content at unprecedented scale

Meta announced it actioned 33.2 million pieces of child sexual exploitation content on Facebook and Instagram globally between January and June 2026, with more than 97% detected proactively before users reported it1

1

. In India specifically, the company took down 5.3 million pieces of child abuse content during the same period, representing approximately 16% of global violations, with a proactive detection rate exceeding 98%2

2

5

5

.

The announcement follows intense regulatory scrutiny after a BBC investigation in July found approximately 30 unique Instagram ads in India promoting child sexual abuse material, with some using explicit terms and linking to Telegram channels selling CSAM for as little as Rs 995

5

. The Washington-based Tech Transparency Project later reported finding 332 ads containing suspected CSAM on Facebook and Instagram in 2026, including dozens targeting India5

5

.

Large language model detection targets signposting ads

Meta introduced new AI tools to combat child exploitation tactics that have evolved to evade detection. The company deployed large language model detection specifically designed to identify "signposting"—seemingly benign ad content suspected of covertly directing users to illegal content or harmful activity hosted on external platforms1

1

2

2

. This addresses a critical gap where ads appear harmless on their surface while serving as gateways to websites hosting child sexual abuse material.

The company fundamentally shifted its ad review systems to examine where advertisements lead, not only what they show. This destination analysis allows Meta to block violating external websites and enforce against the accounts responsible4

4

. Meta emphasized that predators increasingly use advertising to evade detection, creating content specifically designed to bypass automated systems2

2

.

Red-teaming AI agent strengthens platform defenses

Meta deployed a red-teaming AI agent that probes its own defenses for weaknesses, identifying new adversarial tactics before they scale across the platform4

4

. This proactive approach helps the company find methods of abuse before they become widespread. The company also implemented AI-driven sweeps of existing advertisements to surface child sexual exploitation content that earlier systems missed, continuously expanding signals as investigations progress1

1

2

2

.

Recidivism detection systems were strengthened to identify bad actors who attempt to create new accounts after removal4

4

. Meta's broader child safety on Facebook and Instagram relies on multiple technologies including PhotoDNA, which the company has used across all Meta apps since 2011 to detect identical or near-identical copies of known CSAM4

4

. The company maintains one of the industry's largest databases of image and video hashes and shares them with industry partners through the Tech Coalition's Lantern program.

Regulatory scrutiny intensifies in India

Meta faces significant regulatory scrutiny following the BBC investigation. MeitY summoned Meta and on July 4 ordered Instagram to disable all ads and content promoting child sexual exploitation, seeking a detailed explanation within seven days5

5

. Ministry sources indicated Meta could not rely on the "third-party content" defense for paid ads. The National Commission for Protection of Child Rights opened a formal inquiry on August 4 and summoned Meta India's managing director twice in September5

5

.

The National Human Rights Commission directed Delhi Police to investigate whether Meta met its reporting duties under Sections 19 and 20 of the Protection of Children from Sexual Offences Act, 2012, which require reporting such offenses to police5

5

. Delhi Police began registering an FIR against Meta. Previous reports suggested the company could lose safe harbour protections under Section 79 of the Information Technology Act, 2000, forcing it to face greater legal liability for illegal content posted on its platforms3

3

.

Direct reporting to Indian authorities marks operational shift

Meta announced it would report child safety cases directly to the National Cyber Crime Reporting Portal operated by the Indian Cybercrime Coordination Centre2

2

4

4

. This builds on existing reporting of CSAM to law enforcement through the portal with assistance from the National Center for Missing & Exploited Children. While Meta presents direct reporting as its own initiative, the NCPCR stated that its follow-up in the inquiry prompted the change5

5

.

The company's team of experts, including former FBI investigators, conducts manual investigations to proactively uncover, remove and report networks of predatory accounts4

4

. Meta investigates how these networks operate to act at the network level rather than only against individual accounts. The company also introduced parental controls for Meta AI, pre-teen accounts on WhatsApp, and alerts for parents when children search Instagram for self-harm content1

1

. In August, Meta reached an agreement to pay up to $18 billion to settle a child safety lawsuit involving 29 U.S. states1

1

.