Viral TikTok 'Four Emotions' Trend Raises AI Data Collection Concerns

A viral trend on TikTok and Instagram, where users say the same phrase with four different emotional inflections, is raising red flags among digital ethicists. The trend, which involves repeating a sentence in a happy, sad, angry, and neutral tone, has been adopted by millions of users. However, experts like Clara Fulks and Olga Kokhan argue that the trend's standardized format provides an ideal dataset for training AI models on emotional recognition. Large language models (LLMs) like GPT-4o and Claude 3.5 currently struggle to understand nuanced human emotions, making such curated data extremely valuable.

Fulks warns that users may be unknowingly contributing to AI training sets without explicit consent. The trend's structured responses—identical phrasing with varied emotions—allow AI systems to easily map acoustic and linguistic features to specific emotional states. While platforms have not confirmed data collection, the ethical implications are significant: this could accelerate emotional AI capabilities, raising privacy and manipulation concerns. Users are advised to be cautious about participating in trends that yield standardized emotional data, as it may be scraped for commercial or surveillance purposes.