Microsoft has called on the US and other governments to set up their own agencies to regulate AI. Company President Brad Smith is set to deliver a 41-page report on AI’s control, in which he fears the technology will in one way or another erode the stability of society.
Smith said existing government agencies and courts may still apply some laws, they just need experience applying existing laws to AI programs. However, he stressed the lack of containment systems for the smarter, more autonomous AI programs that will emerge over the next decade. Later, new laws and regulations would need to be developed, he said, and a new government agency would be best placed to implement them. Regulating more powerful AI systems requires not only passing new laws, Smith said, but also licensing data centers hosting powerful AI models. A similar idea is known to have been promoted among members of the US Congress by the head of OpenAI, a Microsoft partner, Sam Altman.
Smith emphasized that Microsoft welcomes his call and supports the establishment of a new regulatory authority to implement the licensing system. Such licenses should be required in particular for the use of AI programs that manage the country’s critical infrastructure. In addition, operators must make backup copies of the data. It should also be possible for a human to intervene if the AI starts behaving in a way it shouldn’t.
Smith proposed new rules, stressing that authorities must avoid the mistakes made with the rise of social media. If, in his opinion, a little more than ten years ago, representatives from IT and politics were happy about their role in spreading democracy (he cited the so-called “Arab Spring” as an example), then five years later it became clear that social media as weapon that supposedly aims at democracy itself.
One way or another, the introduction of new laws, standards, and regulations will help Microsoft succeed in AI development, rather than hinder it – the company has ample resources to ensure it meets the high standards set out in the report. Smaller companies will have a much harder time in this case.
However, Microsoft might be cautious even with the tougher AI rules already being developed in the EU. Yesterday, Sam Altham expressed his concern about this – according to him, OpenAI will try to comply with the new law if it is passed, but if this is not possible, it will have to leave the European Union.