SoftBank’s Masayoshi Son says superintelligence in the wrong hands could be “super dangerous”

Masayoshi Son, who has put $64.6bn into OpenAI and calls bubble talk absurd, said in Kyoto that superintelligence in the wrong hands could become super dangerous and that countries must cooperate to control it. He spoke beside Michael Kratsios, who launched a 17-country declaration written in Washington’s renamed vocabulary and endorsed by Germany, Italy, Poland, Greece, Bulgaria, Cyprus and the UK.

Masayoshi Son says he is worried about AI safety. He has put $64.6bn into OpenAI and spent the past year calling bubble talk absurd. He was speaking beside Trump’s science adviser in Kyoto.

Superintelligence in the wrong hands can become ““, Son said on Sunday, as Bloomberg reported. “We don’t have any more luxury of having human fight against human.” Countries have to build trust in order to control the threat, he said.

Bloomberg called the remark uncharacteristic and set it against the security breaches by AI models in recent months. Son has been putting SoftBank money into OpenAI since 2024, and arguing for most of that time that the technology would repay it.

The spending has not slowed.

SoftBank finished paying a $30bn tranche this month, taking its total in OpenAI to $64.6bn and roughly 13% of the company. The final $10bn came out of the largest high-yield corporate bond sale on record, raised a fortnight ago.

Michael Kratsios was in Kyoto for something else. He runs the White House science office, and he arrived with a plan for pointing the technology at scientific research.

Seventeen countries endorsed his Kyoto Vision for a Golden Age of Science, which imagines robotic laboratories running hundreds of experiments at once. The declaration calls the technology super intelligence throughout, five days after the executive order that made it the American government’s word.

Seven of the seventeen are European.

Germany, Italy, Poland, Greece, Bulgaria, Cyprus and the United Kingdom all endorsed it. Six are EU member states, bound by a law called the Artificial Intelligence Act, whose labelling rules began applying in August.

China is not among the endorsers. Argentina, Chile, Indonesia, Kazakhstan and New Zealand are.

Son’s economic forecast was not new. He said in July that AI would be a fifth of global output by 2040, a figure he now puts at $46tn.

The declaration itself says countries endeavour to act, four times over, and never says they pledged or committed to anything. The White House release announcing it says ministers pledged and committed. Nobody who endorsed the text has objected to the stronger description.

Son said the next phase, machines acting in the physical world, begins next year. SoftBank led a $200M round in Zurich’s Gravis Robotics in August, and has been in talks over Munich’s Agile Robots.

Original source SoftBank’s Masayoshi Son says superintelligence in the wrong hands could be “super dangerous”

Back to home