SoftBank Group chief Masayoshi Son, a major AI advocate, warned that superintelligence could become "super dangerous" in the wrong hands as AI capabilities expand exponentially. Speaking in Kyoto on Oct. 4 alongside White House science and tech policy adviser Michael Kratsios, Son said countries must build trust and cooperate to control risks from increasingly powerful AI models. Son has directed nearly US$65 billion of SoftBank funds into ChatGPT creator OpenAI and reiterated his forecast that AI and learning machines could represent at least 20% of global GDP by 2040, equivalent to US$46 trillion. Kratsios announced a US-led initiative involving 17 countries to use AI in scientific research, while China was absent from the nations initially endorsing the White House statement. Son said AI systems could begin helping robots and other machines interact with the physical world from 2027.