A recent TikTok trend has sparked interest and concern, as users repeat the same phrase with varying emotional tones—such as anger, joy, or flirtation. This trend, marked by the hashtag "emotion trend," has gone viral, with over 67 million posts. The practice is believed to help train artificial intelligence (AI) systems to better understand and replicate human emotional expressions through voice modulation. According to Lou Momege of L'essentiel, many social media users may not realize that their participation in this trend could be contributing to the development of AI models. Laurence Devillers, a professor of AI at Sorbonne Universités, suggests that while the origin of the trend is unclear, it could be a strategic effort by a company—possibly based in China—to gather emotional data at no cost. She notes that China has shown a strong interest in emotion recognition technology, which is used in various applications, from customer service to security systems. Devillers explains that similar methods were used in AI emotion detection research in the 2000s, where emotional data was collected to improve AI's ability to interpret human emotions. However, she raises ethical concerns about collecting such data from social media without users' explicit consent. She calls this practice "theft" if users are unaware of how their data is being used. While she acknowledges the value of natural, unscripted data for training AI, she stresses the need for ethical guidelines to protect users' privacy and autonomy. For AI to accurately understand human emotions, it requires more than just vocal tone—it must also consider linguistic and contextual clues. Devillers emphasizes that the complexity of human emotions cannot be captured by simple data collection methods alone. As AI continues to evolve, ensuring that its development respects user rights and ethical standards remains a critical challenge.