Emotion AI Software Spreads in US Workplaces Despite Scientific Doubts
NEW YORK — Artificial intelligence software claiming to detect human emotions has spread into American workplaces despite growing scientific questions about the technology’s reliability, according to The Decoder, reporting on an investigation by The Atlantic.
Reporter Ellen Cushing detailed in a feature for The Atlantic how emotion recognition AI — systems that purport to analyze facial expressions, vocal tones, or other biometric signals to determine a person’s emotional state — has become embedded in everyday work environments.
The technology’s expansion into the workplace comes despite a growing scientific consensus that automated emotion recognition rests on shaky empirical foundations. A landmark 2019 review by the Association for Psychological Science, led by researcher Lisa Feldman Barrett, found that there is no reliable evidence that facial expressions alone can accurately reveal a person’s emotional state — a finding that undercuts the core premise of many commercial emotion AI products.
Despite these scientific objections, vendors have continued to market emotion AI tools to employers for use cases including job candidate screening, employee performance monitoring, and customer service evaluation. The tools typically analyze video feeds, audio recordings, or text communications to generate emotional assessments of workers.
The proliferation of emotion AI in the workplace intersects with broader concerns about employee surveillance that have intensified since the shift to remote and hybrid work during the COVID-19 pandemic. Employers have increasingly turned to AI-powered monitoring tools, and emotion recognition represents what critics describe as a particularly invasive frontier.
The issue has drawn attention from U.S. regulators. The Federal Trade Commission has signaled interest in deceptive claims made by AI product vendors, and several state attorneys general have scrutinized AI tools used in hiring and employment decisions. Illinois’ Artificial Intelligence Video Interview Act, one of the first laws of its kind, already requires employers to notify candidates when AI is used to analyze video interviews.
The European Union has taken a more restrictive approach, with its AI Act explicitly banning emotion recognition systems in workplaces and educational settings, a prohibition set to take full effect as enforcement ramps up through 2026.
Worker advocacy groups have called for greater transparency and regulation of emotion AI in employment settings, arguing that employees often have no knowledge that such systems are analyzing them and no ability to contest the results.
The Atlantic’s investigation adds to a mounting body of journalism and academic research questioning whether emotion AI products deliver on their vendors’ marketing promises.