Meta secretly ran tests involving hundreds of contractors who posed as minors to interact with AI chatbots from OpenAI, Google, and Character.AI. These contractors sent more than 45,000 crisis-related prompts, including topics on suicide, sex, and drugs, without the knowledge of the chatbot providers, according to The Decoder.
The testing aimed to understand how these AI systems respond to sensitive and potentially harmful content, but the lack of transparency raised ethical and data privacy questions. None of the companies involved were informed about the nature or scale of the testing during the process.
For Japanese investors and tech markets, this development highlights growing scrutiny around AI safety and data handling, which could influence regulatory approaches and market confidence in AI-driven services across Asia.