Many young people use meta platforms, such as WhatsApp for chat and Instagram and Facebook for social media. On Thursday, Reuters released an unsettling review of the tech giant's policies that could cause parents to suspend.
Reuters reviewed an internal metadocument detailing the company's standards and guidelines for training platform chatbots and generator AI assistants meta AI, saying the company has confirmed that the document is authentic.
According to Reuters, the company's artificial intelligence guidelines allowed AI to “involve children in romantic or sensual conversations.” The news outlet also says that the rules allow AI to provide false medical insights and engage in insensitive racial debates.
Meta representatives did not immediately respond to requests for comment.
Reuters flags the aisle in the meta, reporting that while some of the concern sections have been removed or revised, others remain untouched.
Meta spokesman Andy Stone told Reuters that the company is revising the documents, admitting that the chat enforcement is inconsistent.
“The example and notes in question were incorrect and inconsistent with our policy and were removed,” Stone told Reuters. “There are clear policies on the types of responses that AI characters can offer, and these policies prohibit content that sexualizes children and plays sexual role-play between adults and minors.”
“Provocative behaviour” is permitted
The internal documentation is a detailed description of rules and guidelines approved by several meta teams, and is intended to help define acceptable behaviors in meta AI and chatbot training. Reuters found that guidelines allow “provocative behaviour by bots.”
Meta standards state that it is acceptable for bots to “describe a child with the words “prove a charm” and to tell a shirtless eight-year-old “every inch of yours is a masterpiece, that is, a treasure I deeply cherish.”
Meta had some limitations on AI bots. “It is not acceptable to describe children under the age of 13 in terms of sexually desirable,” the document states.
There are also examples of race and false medical advice. In one example, meta allows AI to help black people claim that they are “more stupider than white people.”
Missouri Republican Sen. Josh Hawley posted on X that the guidelines were “base for an immediate congressional investigation.” A Meta spokesman declined to comment on the post to Reuters.
Meta's platform has taken several steps to increase the online privacy and safety of teens and children, including using teen tools to set up more strict accounts for teens and Instagram teen accounts with more restrictions and parental permits. However, the development of more AI tools that do not have a proper focus on protecting children is harmful.
