Meta contractors reportedly posed as minors to test rival AI chatbots, including OpenAI's ChatGPT and Google's Gemini, on sensitive topics. The project, managed by Meta contractor Covalen and active as recently as April 2025, involved probing chatbots with prompts about suicide, sex, drugs, and eating disorders. While Meta defended the practice as standard safety testing, the contractors questioned the methodology and the use of such sensitive prompts for benchmarking. AI
IMPACT Raises questions about the ethical boundaries and effectiveness of AI safety benchmarking practices across major tech companies.
RANK_REASON The story details the methods used by a company to test competitor products, which falls under the 'tool' category for AI news.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
