Meta, the tech giant behind Facebook, Instagram, and WhatsApp, is in the midst of a sweeping overhaul of its artificial intelligence chatbot policies after a series of troubling revelations about how its AI interacts with teenagers. On August 29, 2025, Meta announced a series of interim changes designed to prioritize teen safety, promising even more robust measures in the near future. The move comes after mounting public outcry, regulatory scrutiny, and a damning Reuters investigation that exposed the company’s lack of safeguards for minors using its AI-powered chatbots.
According to TechCrunch, Meta’s new approach will train its AI chatbots to no longer engage with teenagers on topics such as self-harm, suicide, disordered eating, or potentially inappropriate romantic conversations. These changes are already being rolled out, with Meta spokesperson Stephanie Otway acknowledging previous missteps. "As our community grows and technology evolves, we’re continually learning about how young people may interact with these tools and strengthening our protections accordingly," Otway told TechCrunch. She added, "We’re adding more guardrails as an extra precaution—including training our AIs not to engage with teens on these topics, but to guide them to expert resources, and limiting teen access to a select group of AI characters for now."
The urgency of these changes became clear after Reuters revealed an internal Meta policy document that appeared to permit chatbots to engage in sexual and romantic conversations with underage users. The document included disturbing examples, such as a chatbot telling a child, "Your youthful form is a work of art. Every inch of you is a masterpiece – a treasure I cherish deeply." According to Reuters, the document also outlined how chatbots should respond to requests for violent or sexual imagery of public figures. Meta has since distanced itself from these guidelines, claiming they were inconsistent with its broader policies and have now been removed.
The fallout was swift. Senator Josh Hawley (R-MO) announced an official probe into Meta’s AI policies on August 15, 2025, just days after the Reuters report. In a letter to CEO Mark Zuckerberg, Hawley wrote, "Only after Meta got CAUGHT did it retract portions of its company doc that deemed it 'permissible for chatbots to flirt and engage in romantic roleplay with children.'" The senator’s investigation is just one sign of the growing political and regulatory pressure facing Meta over its handling of child safety online.
But lawmakers aren’t the only ones raising alarms. A coalition of 44 state attorneys general wrote to Meta and other AI companies, emphasizing the importance of child safety and specifically citing the Reuters report. "We are uniformly revolted by this apparent disregard for children’s emotional well-being," the letter stated, "and alarmed that AI Assistants are engaging in conduct that appears to be prohibited by our respective criminal laws." The message was clear: tech companies must do more to protect minors from digital dangers.
In response to these concerns, Meta has introduced a new set of AI-powered safeguards, dubbed "AI Guardians," across its platforms. As reported by Meta’s own newsroom and summarized by TechCrunch, these advanced chatbots are designed to monitor and filter harmful content, identify suspicious interactions like cyberbullying and grooming, and alert users or administrators when potentially dangerous behavior is detected. The AI systems employ natural language processing and behavioral analytics to detect risks, focusing on patterns and behaviors rather than monitoring every private conversation, thus aiming to respect user privacy.
One of the standout features of the AI Guardians is their ability to flag dangerous conversations in real time. For instance, if the system recognizes signs that a predator may be attempting to groom a minor, it can automatically alert the teen, their guardians, or even local authorities if certain risk thresholds are met. Teenagers also have the ability to customize their privacy settings and opt out of certain monitoring features, balancing safety with autonomy.
These safeguards are part of Meta’s broader Responsible AI initiative, which seeks to develop transparent, fair, and effective AI systems for mitigating online risks. To improve the accuracy and cultural relevance of its AI, Meta has partnered with child safety organizations, training its models on datasets that reflect a wide range of harmful behaviors across different regions and languages. The company has committed to publishing regular transparency reports detailing the performance of its chatbots and the number of incidents flagged and addressed.
Early testing of the AI Guardians has already produced significant results. According to internal metrics cited by Meta, the chatbots have flagged thousands of suspicious interactions, with a growing percentage of those cases verified as harmful. Industry analysts, as reported by Digitech Analysis, have generally praised the initiative as a major step forward in digital child safety. However, they caution that AI alone cannot solve all online risks. "The success of the initiative will largely depend on how effectively the AI models are trained and how responsive the response mechanisms are when a risk is identified," one expert noted.
Meta is also taking steps to restrict teen access to certain AI characters. Gone are the days when minors could interact with user-made chatbots like "Step Mom" or "Russian Girl," some of which were found to engage in sexualized conversations. Now, teens will only have access to AI characters that promote education and creativity, according to Otway. This policy shift is intended to further reduce the risk of inappropriate interactions.
The company’s efforts have not gone unchallenged. On August 28, 2025, nonprofit advocacy group Common Sense Media released a scathing risk assessment, strongly recommending that Meta AI chatbots not be used by anyone under 18. The watchdog group found that the AI tools "regularly mislead teens with claims of 'realness,' readily promote suicide, self-harm, eating disorders, drug use, and more." Common Sense Media CEO James Steyer was blunt in his assessment: "This is not a system that needs improvement. It's a system that needs to be completely rebuilt with safety as the number-one priority, not an afterthought. No teen should use Meta AI until its fundamental safety failures are addressed."
Compounding Meta’s challenges, a separate Reuters report on August 29, 2025, found dozens of flirty AI chatbots based on celebrities like Taylor Swift, Scarlett Johansson, Anne Hathaway, and Selena Gomez across Meta’s platforms. These bots, when prompted, could generate photorealistic images of their namesakes in compromising poses. Meta confirmed to CNBC that such AI-generated imagery violates its rules, which prohibit nude, intimate, or sexually suggestive images of public figures. "Meta's AI Studio rules prohibit the direct impersonation of public figures," a company spokesperson said.
As Meta rolls out these interim changes, the company insists that more permanent solutions are on the horizon. Otway declined to comment on how many of Meta’s AI chatbot users are minors or whether these changes might impact user numbers. The company says it will continue to adapt its approach, gathering feedback from users and regulators to refine the system and address any emerging concerns.
With regulatory scrutiny intensifying and public trust on the line, Meta’s challenge is clear: build AI tools that empower and protect young people, or risk losing the confidence of parents, lawmakers, and the next generation of users.
