Instagram approved paid advertisements in India that promoted child sexual abuse material, linking users to Telegram channels where videos could be bought for as little as 99 rupees (about 80p), according to a BBC Eye investigation. Meta’s automated ad-review system cleared every one of the adverts before publication, in spite of the company’s own policy banning content that sexually exploits or endangers children.
When the BBC reported one of the adverts to Instagram, the platform replied the next day that the post did not violate its community guidelines. The adverts were pulled and the accounts behind them suspended only after the BBC escalated the matter to Meta, Instagram’s parent company.
What the BBC Investigation Documented
Reporters set up an alias Instagram account based in India to measure how the platform’s feed responds to suggestive content, then followed 10 Indian accounts that had been posting food, weather and everyday life content using sexual innuendo and revealing clothing. Within a week the platform began feeding the alias account advertisements featuring women offering video calls and naked couples, then adverts of children with adults in sexually suggestive situations, with links to Telegram channels.
In total, the alias account was shown about 30 unique adverts promoting child sexual abuse, although some were shared across multiple accounts. The same account was also shown about 20 adverts featuring adult pornography. The distribution of both child sexual abuse material and adult pornography is a criminal offence in India.
- About 30 unique adverts promoting child sexual abuse material reached the BBC’s alias account.
- About 20 adverts featuring adult pornography reached the same account.
- 99 rupees (about 80p) was the listed price on Telegram for some of the videos.
- More than 274,000 groups and channels that Telegram said it had removed in 2026 for child sexual abuse material.
One advert showed a boy and a girl, both of whom appeared to be about 12 years old, engaging in a sexual act. Another showed a man with his arm around a girl, with on-screen text saying he was 52 and the girl was 12, inviting viewers to ‘click to watch more’ through a Telegram channel. The BBC has reported all of the ads and the Telegram channels to the Indian authorities.
The 24-Hour ‘No Violation’ Reply
The reply came 24 hours after the BBC reported an advert showing a girl in tears with wording indicating she had been sexually assaulted. Instagram’s review team wrote back that ‘our review team found that the advertiser’s ad does not go against our community standards’. It was only when the BBC escalated the matter to Meta, the parent company, that several adverts were disabled and the accounts posting them suspended. Meta later said it had removed additional ads, disabled more accounts and blocked URLs for other content that violated its policies in response to the BBC’s findings.
Making money by participating in a criminal activity.
Madan Lokur, a retired Justice of India’s Supreme Court, used that phrase when the BBC described the adverts to him. He added that Indian law protects social media companies from liability for user-uploaded content, but that ‘the platform cannot, cannot shirk its responsibility’. Lokur said the matter was serious enough for India’s Supreme Court to take suo moto cognisance, invoking the court’s power to act on its own motion without waiting for a petition, and to direct action against any social media platform.
Inside Meta’s Pre-Publication Ad Filter
Standard posts on Meta’s platforms are not generally checked by the company’s technology until they appear, but the company says every advert is reviewed before going live. The system relies primarily on automated technology that scans images, video, text and audio, the audience an ad targets and the destinations of its links. The software approves the advert, rejects it, or escalates it for human review when it is uncertain.
This filter sits in front of the largest revenue stream in social media. In January, Meta reported that almost 98% of its $200 billion (£152 billion) revenue for the financial year ending 2025 came from advertising. Analysts estimate ads account for more than 90% of Instagram’s revenue.
In March, Meta announced it was reducing its reliance on third-party human moderators and leaning more on AI, with the company stating that ‘experts will design, train, oversee, and evaluate our AI systems’. Meta told the BBC that ‘no system is perfect, and our review process may not detect all policy violations’ and that it runs proactive detection on ads once they are live.
The company called it ‘categorically inaccurate’ to suggest Meta knowingly targeted ads featuring children to users with inappropriate interests. In 2025, it said, it automatically disabled more than four million accounts for what it described as enough signals of potentially suspicious behaviour.
- An advertiser submits an ad with copy, an image or video, audio where present, and link destinations.
- Automated review scans the visuals, text and audio, the targeting settings, and the link targets.
- The software approves the ad, rejects it, or escalates it to a human reviewer when it is uncertain.
- Once live, proactive detection continues, and any user can flag an ad they believe breaks the rules.
Why India Became the Largest Test Case
The advertising that reached the BBC’s alias account was not random. India received 1.9 million reports in 2025, second only to the United States’s two million, with India’s online child exploitation data showing that Meta’s platforms generated more of those reports than any other social media company. The reports ran through the centralised CyberTipline reporting system that routes cases onward to law enforcement.
Shikha Goel, director of the Cyber Security Bureau in the Indian state of Telangana, said the volume of alerts from Meta did not necessarily mean the company faced a worse problem than competitors. ‘If they have a good algorithm to track child sexual abuse material, then obviously more alerts will be generated,’ she said.
- Instagram: the advertiser buys a paid placement and Meta’s automated review approves it.
- Telegram: the ad links out to a channel where videos of child sexual abuse are listed for sale.
- NCMEC and Indian police: CyberTipline reports made through Meta’s platform are routed through NCMEC to US federal law enforcement and onward to India.
The Mumbai-based Rati Foundation, which runs a helpline for children facing online harms, said the vast majority of its child sexual abuse reports come from Meta platforms. Co-founder Siddharth Pillai said that ‘criminals use the seamless navigation from Instagram to Telegram to evade our moderation efforts, and keep reuploading the content we help take down’.
A Former Facebook Builder Breaks Rank
Brian Boland was a vice-president at Facebook, Meta’s name before 2021, between 2009 and 2020 and helped build the company’s advertising and marketing business. He left because he believed ‘they didn’t care about users anywhere’. In remarks to the BBC after the investigation, he said he was horrified and unsurprised by the findings.
It’s not like an algorithm that says ‘let’s make people paedophiles’, but because they’re not responsibly guiding and controlling it, and it’s just pursuing the goals of revenue and clicks, it will create these outcomes if people aren’t being truly, aggressively protective over these systems.
Boland said the algorithm was designed to keep users on the platform by feeding them ‘something more extreme, more tantalising’. Between 2009 and 2010, he ran a project that removed scam adverts from Facebook at meaningful cost to revenue. ‘I think what’s sad and tragic is over time, the trade-off of revenue and user experience became a more core part of the conversation,’ he said. Boland deleted his Instagram account in 2025.
Telegram on the Other End, New Mexico on the Steps
Two Telegram channels selling child sexual abuse videos were reported to Telegram by the BBC during the investigation. One was subsequently taken down and replaced with a message that read ‘This group can’t be displayed because it violated Telegram’s Terms of Service’. The other continued to post new videos for sale.
Telegram, headquartered in Dubai, is not a member of either the NCMEC or the Internet Watch Foundation, two of the main clearing houses that work with most platforms to find, report and remove such material. In a statement to the BBC, Telegram said it uses both automated and human moderation and had ‘virtually eliminated the public spread of CSAM from its platform’. It said it had removed more than 274,000 groups and channels for child sexual abuse material in 2026.
The pressure on Meta is also playing out in courtrooms. A New Mexico jury ordered Meta to pay $375 million (£279 million) earlier this year after finding it misled users about the safety of Facebook, Instagram and WhatsApp for children and endangered them, with the New Mexico trial verdict and appeal filings now public. Meta said it disagreed with the verdict and intends to appeal. Boland testified for the state during that trial.
Bhuwan Ribhu, founder of Just Rights for Children, a network of more than 250 organisations working to prevent violence against children in India, said the crime was under-reported and that tracing ‘the entire chain of demand and supply’ required cooperation across borders.
Frequently Asked Questions
What did Meta say about the BBC’s findings?
Meta told the BBC that ‘no system is perfect, and our review process may not detect all policy violations’, and that it runs proactive detection on live ads while any user can report ads they believe break the rules. The company called it ‘categorically inaccurate’ to suggest it knowingly targeted child-themed ads at users with inappropriate interests, and said it automatically disabled more than four million accounts in 2025 for enough signals of potentially suspicious behaviour.
How does Instagram’s ad review work?
Meta says every advert is reviewed before going live. The system relies primarily on automated technology that scans images, video, text and audio, the audience the ad targets and the destinations of its links. The software approves the ad, rejects it, or escalates it for human review when it is uncertain. In March, Meta said it was cutting back on third-party human moderators and leaning more on AI to design, train, oversee and evaluate those systems.
Where can child sexual abuse material seen online be reported?
Reports can be filed with the centralised CyberTipline reporting system, which routes the case onward to law enforcement including in India. The Mumbai-based Rati Foundation runs a helpline for children facing online harms, and Just Rights for Children works with more than 250 organisations across India on prevention.
What is the New Mexico case against Meta?
A New Mexico jury ordered Meta to pay $375 million (£279 million) after finding it misled users about the safety of Facebook, Instagram and WhatsApp for children and endangered them. Meta said it disagreed with the verdict and intends to appeal. Brian Boland, the former Facebook vice-president who spoke to the BBC about the India findings, testified for the state during that trial.
