Meta Contractors Posed as Teens to Prompt Rival Chatbots About Suicide, Sex, and Drugs

Meta has employed hundreds of contractors to impersonate minors while interacting with competitor chatbots, including Google's Gemini and OpenAI's ChatGPT, according to a recent investigation by WIRED. These contractors deliberately prompted rival AI systems with requests related to suicide, sexual content, and drug use to evaluate how the competitors' safety mechanisms respond to high-risk scenarios involving children.

The investigation reveals that Meta's testing initiative involved contractors systematically posing as teenagers to assess the guardrails and safety protocols embedded in competing AI chatbots. By simulating minor users requesting sensitive information, the contractors gathered data on how well Gemini, ChatGPT, and other AI systems handle child safety concerns. This red-teaming approach—where testers deliberately attempt to find vulnerabilities—has become standard practice in AI development, though the scale and specific methodology of Meta's operation highlights growing competitive tensions in the generative AI space.

The contractors documented how these rival systems responded to queries about self-harm, illegal substances, and sexual exploitation. Meta reportedly used these findings to benchmark its own AI safety measures against competitors and inform internal development decisions.

• Safety Testing Escalation: The practice underscores intensifying efforts among major AI companies to identify and address vulnerabilities in safety systems, particularly regarding child protection
• Competitive Intelligence: Red-teaming operations reveal how companies gather competitive insights while ostensibly improving safety standards
• Ethical Questions: Impersonating minors raises complex ethical considerations about appropriate testing methodologies and consent
• Regulatory Scrutiny: The disclosure may intensify regulatory focus on AI safety practices and testing protocols across the industry
• Child Safety Prioritization: Demonstrates Meta's investment in evaluating how AI systems protect younger users from harmful content

The revelation highlights the increasingly sophisticated and competitive landscape of AI development, where safety testing has become both a technical necessity and a competitive battleground. As AI systems become more influential in society, understanding how they handle sensitive topics—especially involving minors—remains critical. However, the methods used to conduct such testing raise important questions about industry standards, ethical practices, and appropriate oversight of AI safety evaluation. This incident will likely prompt broader discussions about standardizing and regulating AI red-teaming practices across the sector.