A young girl holding smartphone while lying on floor in cozy home interiorgettyDespite Meta’s investments in AI and its continued growth, Instagram’s own AI-driven ad review system is still allowing advertisements that promote child sexual abuse material (CSAM)—at least in India. When one journalist flagged it, the system replied that it did not break the rules.The evidence comes from one example but the story is clear: A company that can out-invest most nations in artificial intelligence still cannot keep child exploitation off its own platform. The reason, as industry experts have remarked, has less to do with capability than with what Meta is actually optimizing for.The Evidence of Meta’s Error, By The NumbersThree separate investigations over the past year point to the same underlying pattern: Meta's safety failures track its revenue incentives, not its technical limits.July 6, 2026: India's IT Ministry ordered Meta to disable the ads within days of the BBC's report and explain within seven days how they passed review, putting the company's legal safe harbor in the country at risk.July 3, 2026: A BBC Eye investigation found a test Instagram account in India was served 30 ads using terms like "rape video" and "child video," linking to Telegram channels selling the material for 99 rupees (about $1). When BBC reported one ad, showing a crying child with text suggesting sexual assault, Instagram's review team said it "does not go against our community standards."March 24, 2026: A Santa Fe jury found Meta liable for misleading the public about child safety on Instagram and Facebook, ordering $375 million in damages. Former Meta vice president Brian Boland testified that Instagram's algorithm was built to show users "something more extreme, more tantalising," and told the BBC he left the company because "they didn't care about users anywhere."November 6, 2025: Reuters documents revealed Meta projected roughly 10% of its 2024 revenue, about $16 billion, would come from ads for scams and banned goods. The company bans advertisers only when automated systems are 95% certain of fraud; below that threshold, it charges "penalty" rates instead of removing them. Internal targets aim to cut that share to just 5.8% by 2027, not eliminate it.Two figures, Meta documented in their January investor update, anchor the contrast: Meta earned $201 billion in 2025, with 97% from advertising, while spending $72.2 billion on AI infrastructure that year. It has now raised 2026 capital spending guidance to $125-145 billion, an 87% jump, specifically to build out the AI systems meant to run its platforms and review its ads. What It Means For Meta And Industry LeadersThe BBC’s test account didn’t hide anything. It followed 10 public accounts and Instagram’s own systems began surfacing sexually suggestive ads within days, then child exploitation ads within a week, without the user searching for either. The system that failed here is the same type of system Meta and other industry leaders are spending tens of billions to expand.MORE FOR YOUHow much longer will the public allow such behavior before taking issue with their representatives and looking for alternatives? This is the question industry leaders need to keep an eye on. Based on recent lawsuits and court decisions, it appears that the public fervor is growing rapidly and meaningful action from governing bodies is closer than ever before.