Brad Smith urges safety testing, disclosure rules and the ability to switch AI systems off
Microsoft President Brad Smith has renewed his call for tougher AI oversight, telling ITV News that AI is currently “less regulated than a carton of milk.”and Smith said the time to act is now.
What he wants
According to ITV’s report, Smith wants three things:
Disclosure: Companies should be required to report when AI agents go rogue or when users are hacked.
Testing: Independent evaluations and safety testing should happen before models are launched.
Off-switch: There should be a way to switch AI systems off. Separate reporting says Smith has backed the idea of an AI “kill switch,” arguing that the people who put controls around AI agents should be able to turn them off.
Jobs and the economy
Smith also expects jobs to be lost as AI is rolled out, and says governments will need to step in to help people through the transition. He pointed to skills training, taxing companies that use AI or robots to replace human labour, and possibly protecting certain roles, such as teachers, doctors, judges and care workers, from being performed by AI.
A familiar argument
The milk comparison is not new. In 2024, Smith told Euronews that if AI is the most powerful technology in the world, it should have at least as much regulation as a carton of milk. His reasoning was that shoppers buy milk without worrying about safety because a regulatory floor exists. He has also long backed “safety brakes” for AI and, in 2023, called for a new US agency and licensing for the most powerful AI tools.
Political backdrop
ITV notes that US President Donald Trump has called concerns about AI’s dangers a hoax and has stressed the importance of America winning the AI race with China.