SoftBank Founder Masayoshi Son Warns of AI Safety Risks and Urges Global Cooperation
Speaking alongside White House technology advisor Michael Kratsios at a forum in Kyoto, SoftBank founder Masayoshi Son issued a rare warning that the exponential growth of AI capabilities requires urgent international cooperation to prevent superintelligence from falling into dangerous hands. As one of the world's most prominent AI bulls and a major backer of OpenAI, Son's shift toward public caution underscores growing anxiety within the tech ecosystem regarding security vulnerabilities and uncontrolled AI risks. His commentary highlights the escalating need for unified global AI governance as models advance rapidly. Son outlined that AI is entering a 'third phase' starting next year, where models will directly empower physical robotics and hardware to interact with the real world. He also reiterated his forecast that AI and autonomous machines will contribute at least 20% of global GDP ($46 trillion) by 2040.
## BACKGROUND
Artificial Superintelligence (ASI) refers to hypothetical AI models that surpass human intelligence across virtually all economically valuable tasks, raising existential and catastrophic safety concerns among researchers. Meanwhile, embodied AI (or Physical AI) integrates these advanced machine learning models into physical hardware and robotics, enabling software systems to execute actions in the physical environment.