Meta has been exposed for allowing advertisements promoting child sexual abuse material in India, and local authorities have asked the social media giant to explain why its ad review system failed to detect these illegal promotional content in a timely manner.

According to BBC reports, multiple versions of advertisements promoting child abuse content have appeared on Meta's apps in recent weeks. These ads, seen by the BBC World Service, used keywords such as "rape video" and "child video" and directed users to channels on the messaging app Telegram, where users could purchase such material for as little as 99 rupees (about $1).

The BBC said it reported these violating ads to Meta but was initially told they did not breach its guidelines. Meta subsequently removed the promotional content.

This report comes as Meta continues to push forward with plans to replace most human moderators with AI agents. These automated systems are authorized to assess and handle potentially violating content on its platforms.

Meta did not specify whether these violating ads were caused by its AI processes, but told the BBC: "No system is perfect, and our review processes may not detect all policy violations."

This incident is another blow to Meta's ambitions to expand AI applications. The company hopes to eventually carve out a significant additional revenue stream by offering AI automation services to third-party customers.

Meta has cut about 20% of its workforce this year, with a considerable portion of those roles replaced by AI agents. The company appears to be using its own operations as an example of how its AI tools can reduce labor costs.

However, these early issues with its AI systems could become a significant obstacle to its development.

Last month, more than 20,000 Instagram accounts were compromised after a group of users discovered they could convince Meta's account support agent to grant access without providing any proof that they were the actual account holders. Meta said it has patched the issue, but the scale and scope of how people query AI chatbots seems to have enabled nearly limitless exploitation of such vulnerabilities.

Meta also faces challenges in AI development. Reuters reported last week that Meta CEO Mark Zuckerberg told employees that the development of its AI agents was not meeting expectations, noting that recent layoffs had not yet yielded the planned benefits. The Information reported that Meta has also imposed restrictions on AI token usage due to soaring costs from its forced reliance on AI tools.

These costs may be why Meta is currently exploring new ways to profit from its massive AI investments, including developing cloud infrastructure products to maximize utilization of its ever-expanding data center network.

The additional data capacity was initially reserved for Meta's own AI use, but it now appears the company is reassessing its overall AI advancement strategy, and even the true value of its potential AI products. Issues with its internal AI use may further heighten concerns.

Meanwhile, Meta faces other public relations issues in the AI space. Last week, Wired reported that Meta deployed a large number of contractors posing as minors to test online how competitor chatbots respond to prompts involving suicide, sex, eating disorders, and other high-risk topics. Meta said this was standard research to better understand safety guardrails and mitigations. But from a perception standpoint, it is another blow to its AI push, which already faces significant public opposition.

If Meta's AI systems fail to protect children, this could be the last straw for its approach. It could also have a major impact on the company's business future, especially if Meta fails to build more advanced AI models through its highly touted superintelligence team.