The Microsoft chairman is calling for ‘emergency brakes’, mandatory incident reporting and an independent assessment of models, and sees the UK taking the lead on these rules.
Guest of podcast Big Boss Interview Brad Smith, President of Microsoft, said a washing machine or a milk brick is now subject to more security checks than artificial intelligence. He calls for international rules, breaking the line of Donald Trump, who prefers to let the sector self-regulate so as not to give way to China.
Incident reports and public evaluators
Among its proposals are: emergency arrest mechanisms to suspend an out-of-control system, the legal obligation to write incident reports and prevent victims when an AI agent "fails" and compromises their systems, as well as a review of new models by independent government evaluators prior to placing them on the market.
Brad Smith believes that the UK is best placed to wear these rules through its AI Security Institute, and believes that Washington would be more willing to accept standards from London than Brussels. He calls AI both a tool and a weapon, and considers that the time has come to act on agents beyond human control and models threatening global cybersecurity.
Read the full article on the original source
Go to the source →