On October 26, 2026, MIT Technology Review published a commentary titled “Focusing on AI Safety Disasters While Masking Everyday Harm,” criticizing current tech policies for narrowly defining “success” as preventing extreme incidents—such as AI chatbots discouraging suicide or social media platforms blocking children from purchasing fentanyl—while tolerating the ongoing, hidden yet widespread societal harm caused by these products.
The article points out that teenagers develop pathological attachments to AI companions, students rely on chatbots to write entire papers, and young people have their sense of self-worth distorted by algorithmically curated, jealousy-inducing information feeds. Though such phenomena rarely make it into court cases or headline news, they are systematically eroding cognitive habits, learning abilities, and genuine interpersonal relationships. MIT researchers have found that adolescents who use ChatGPT to write papers exhibit reduced brain connectivity during the task, with the “cognitive debt” effect persisting even after the task is completed.
The author reveals how tech giants shape the regulatory agenda through lobbying: In 2026, Google supported multi-state AI chatbot legislation that mandates only basic safeguards—such as blocking harmful content and preventing self-harm inducement—while including broad exemption clauses that effectively allow mainstream services to sidestep these requirements. This industry-driven legislation has successfully taken effect in places like Nebraska, whereas stronger proposals aimed at preventing children’s emotional dependence on AI have been shelved.
The article draws on historical lessons to underscore the persistence of risks: Meta’s platform has been accused of exacerbating violence in Myanmar’s Rohingya genocide, Sri Lanka’s anti-Muslim riots, and Ethiopia’s Tigray war; now, AI is extending this same logic into deeper aspects of daily life—nearly one-third of teenagers choose to confide in AI about major life issues, as technology increasingly replaces the cognitive training and emotional development that should ideally be fostered through genuine human interaction.
At the end of the article, three specific demands are put forward: educational institutions and technology companies must provide concrete evidence demonstrating that AI enhances learning and critical thinking before it can be introduced into classrooms; AI companions designed to simulate consciousness or express affection must be prohibited; and independent reviews are required to assess the actual impact of AI systems on interpersonal relationships and the capacity for autonomous thought. The author emphasizes that technology companies, which have become deeply embedded in modern life, bear responsibilities that extend far beyond merely avoiding the next shocking headline.