SoftBank’s Son warns of AI safety risks despite $65B OpenAI bet

SoftBank Group (SFTBY) (SFTBF) Chief Executive Masayoshi Son warned that rapidly advancing artificial intelligence could pose serious safety risks, an unusual note of caution from one of the technology's most aggressive investors, Bloomberg News reported.

Speaking Sunday at an event in Kyoto alongside White House science and technology policy adviser Michael Kratsios, Son called for greater international cooperation to manage increasingly powerful AI systems. He warned that superintelligence could become dangerous if controlled by the wrong actors as AI capabilities continue to expand.

The comments are notable because SoftBank has committed nearly $65 billion to OpenAI (OPENAI), making Son one of the biggest financial backers of the AI boom. He has repeatedly argued that artificial intelligence could transform the global economy and generate enormous investment opportunities.

Son's warning highlights a growing tension surrounding the AI investment boom. Companies are committing hundreds of billions of dollars to chips, data centers and advanced models while governments are considering how to manage security and safety risks. Tighter oversight could raise costs or slow deployment, while continued rapid development would support demand across the AI infrastructure supply chain. For SoftBank investors, the stakes are particularly high because of the company's enormous financial exposure to OpenAI and Son's broader AI strategy.

Son nevertheless remains strongly bullish on the technology's economic potential. He expects AI to move beyond chatbots and coding tools toward systems that control robots and interact directly with the physical world, with that transition beginning as soon as next year.

He also reiterated his forecast that AI and intelligent machines could account for at least 20% of global gross domestic product by 2040, representing about $46 trillion in economic activity.

Kratsios, meanwhile, announced a U.S.-led initiative involving 17 countries aimed at using AI to accelerate scientific research. Participating nations include Japan, South Korea, the U.K., Germany, Singapore and the United Arab Emirates, while China was not among the initial signatories.

The initiative envisions AI agents analyzing scientific literature and automated laboratories conducting large numbers of experiments simultaneously. The effort illustrates the dual challenge facing governments and investors: capturing the potential productivity gains from increasingly capable AI while addressing the security risks that could accompany them.