India’s Ministry of Electronics and Information Technology has given Meta seven days to explain how paid advertisements on Instagram, some allegedly containing explicit search terms and directing users to Telegram channels selling child sexual abuse material for less than the price of a cup of tea, passed through the company’s own advertising review systems. Meta’s response, a familiar refrain about a zero-tolerance policy and a “constant battle” against criminals hiding among 3.5 billion users, is not really the point.
The point is that nobody hid anything. An investigation reportedly built a fresh account, followed a handful of accounts posting merely suggestive content, and within days was being served advertisements for material depicting the sexual abuse of children. Nothing was searched for. The platform’s own recommendation system did the finding. That distinction matters more than most of the coverage has allowed. When Instagram approves an advertisement, it does not simply permit it to exist. It reviews the creative, prices the placement, selects the audience, and pushes it into a specific feed.
That is a commercial act performed by the platform itself, not a passive act of hosting someone else’s speech. Every other advertiser on Instagram accepts that the platform bears responsibility for what it approves and profits from. That responsibility does not evaporate the moment the product being sold is access to a child’s exploitation. Safe harbour, the legal shield that protects intermediaries from liability for content they host but did not create, exists because a platform cannot plausibly review billions of daily uploads before they appear.
It was never designed to cover a paid placement that the platform’s own systems screened, priced, and distributed. India’s IT Rules already require large platforms to deploy automated detection for exploitative content within compressed timelines, so the principle that an auditable child safety standard is achievable is already accepted in law. What has not happened is extending that same standard, with published detection figures and independent audits rather than self-certified statements no outsider can verify, to the advertising pipeline specifically.
There is a second problem the current notice does not fully address, and it is arguably more troubling than any single failure to screen an advertisement. Recommendation systems are built to maximise engagement, and by the investigation’s account, they moved a test account from legal but suggestive content toward advertisements for illegal material in the course of ordinary use, without any deliberate search.
If that account of the mechanism holds up, this is not merely a moderation failure in the sense of a bad actor slipping past a filter. It is a system that, in the process of doing exactly what it was built to do, constructed a pathway toward criminal content. A platform can fix individual advertisements one at a time for years and still leave that underlying mechanism untouched. Any explanation Meta submits should be judged on whether it addresses the recommendation engine itself, no t only the advertisements it happened to serve up.
There is also a credibility question the ministry has not had to answer before, but will now. India has sent stern notices to major platforms over harmful content before, including to X, YouTube, and Telegram, without any of them ultimately losing safe harbour protection as a result. A notice that ends in a letter changes very little. If this episode follows the same script, a public statement, an internal review promised within seven days, and then quiet closure, the deterrent effect on every other platform’s advertising pipeline will be close to zero.
What would change the calculation is a published finding that ties a specific, revenue generating failure to a specific legal consequence, whether that is a fine, a referral for prosecution under the Protection of Children from Sexual Offences Act, or an explicit, tested finding that safe harbour did not apply to that placement. A third gap sits outside this particular notice but belongs in the same conversation. India generated close to 1.9 million reports of online child exploitation last year, second only to the United States, yet nearly all of them reach Indian authorities only because American platforms are required by American law to route them through a tipline based in the United States.
India has no domestic law compelling platforms to report this material directly to Indian agencies, and none at all requiring disclosure of whether the material surfaced through an ordinary post or a paid, reviewed advertisement. Whether Meta’s advertising failures in this case ever reach an Indian regulator through anything other than a notice like this one currently depends on a foreign statute and the company’s own discretion. None of this requires reinventing platform law from scratch.
It requires three specific moves. First, treat a paid, approved advertisement as forfeiting safe harbour for that placement, since approval of an advertisement is a commercial decision the platform makes, not a passive act of hosting. Second, enact a domestic mandatory reporting law for child exploitation material, with disclosure duties that apply specifically to advertising and recommendation systems. Third, require the same automated, independently audited detection standard that protects a platform’s own revenue from fraud to be applied, and verified by someone outside the company, to the pipeline that is supposed to protect children.
The seven-day clock the ministry has set will run out, and Meta will send its explanation. A more consequential clock is the one measuring whether India, this time, turns a notice into a legal standard that outlasts the news cycle, or whether an algorithm that sold access to a child’s exploitation is treated, once again, as a moderation failure to be explained away rather than a business practice to be made illegal.
(The writer is a tech and social entrepreneur and Programme Director, Eastern India, at WHEELS Global Foundation. He writes on AI, economy and geopolitics and can be found on X at @ipravinkaushal.)