Akira Hayashi is dead serious when he says artificial intelligence could kill us in the near future.

The 20-year-old Tokyo native didn¡¯t become an AI ¡°doomer¡± overnight. Four years ago, while still in high school, he picked up Nick Bostrom¡¯s 2014 book ¡°Superintelligence: Paths, Dangers, Strategies,¡± initially expecting it to be little more than a collection of conspiracy theories. As he read on, however, he found its arguments more difficult to dismiss than he had anticipated. He began poring over academic papers and internet forums devoted to the potential dangers of AI. After enrolling at the University of Tokyo, he established a community focused on AI safety.

His concerns deepened in 2025 with the online publication of ¡°,¡± a report examining the exponential pace of AI development. It outlined a scenario in which, unless humans act by 2027, AI systems surpassing human intelligence could begin deceiving people and pursuing goals that threaten humanity¡¯s survival, potentially through cyberwarfare or the release of biological weapons.