Microsoft’s president stated in an interview with POLITICO Tech that the business needs federal licences for any AI that is potent enough to pose a threat to public safety.
In a Senate subcommittee hearing yesterday and in an interview, Microsoft President Brad Smith supported the idea of a federal licencing regime and a new agency to regulate potent AI platforms. This position puts Microsoft at odds with Google and some other tech giants, who prefer voluntary guidelines with light penalties for those who develop AI technology.
Smith, Elon Musk of Tesla, Mark Zuckerberg of Meta, Sam Altman of OpenAI, and Satya Nadella, also of Microsoft, are expected to attend a closed-door meeting with Senate Majority Leader Chuck Schumer on Wednesday as Congress conducts a frenzy of AI hearings.
According to Smith, the main goal of a licencing system is to guarantee a specific standard of competence and safety. Before we can receive a licence, we must demonstrate that we can drive. We risk losing control if we drive carelessly. You can employ the similar ideas, especially for applications of AI that have safety implications.
Sens. Richard Blumenthal (D-Conn.) and Josh Hawley (R-Mo.) have suggested a legislative framework that would establish a licencing body for complex or possibly hazardous AI models. Smith testified in favour of this framework on Tuesday in the Senate. The framework also demands that businesses be held responsible when their AI models violate civil rights, damage privacy, or result in other cognizable harms.
According to the Microsoft executive, only dangerous AI systems should be subject to licencing requirements, and the process shouldn’t be too difficult, time-consuming, or expensive. However, he anticipates both Microsoft and OpenAI, in which Microsoft has a significant stake, would be bound by the licencing laws. The business is ready to put its words into action.
He told POLITICO Tech that it is a reasonable strategy that Congress should adopt as a “first step” and immediately add to.
According to Smith, our sector has taken a while to understand that, if applied thoughtfully and sensibly, law and regulation may improve a market. But now, when he interacts to folks in the tech industry, they are thinking, let’s just say, with maybe a little more experience and possibly even a little more wisdom than we had ten years ago.
Smith agreed that Congress was unlikely to enact any significant AI legislation this year, but he expressed optimism that regulation will ultimately arrive — potentially next year or “beyond.”
Smith believes that other items that could pose a risk to consumer safety have long been subject to regulation, including food, prescription pharmaceuticals, and motor vehicles. He believes more tech leaders now understand that artificial intelligence will also need to be subject to regulation.
Smith stated that industry should be heavily involved in formulating the guidelines for AI regulation.
One of the seven businesses that signed the White House’s voluntary safety standards earlier this year was Microsoft. On Tuesday, eight additional IT firms joined the pledge, including Adobe, IBM, and Palantir. According to Smith, the procedure can serve as a template for other government bodies looking to regulate AI.
Following a meeting in May, the White House provided Microsoft, Google, Anthropic, and OpenAI with a month to provide their own rules, Smith claimed. After reviewing their pitches, administration representatives advised them of any additional work that was required. Smith stated on POLITICO Tech that Gina Raimondo, the secretary of commerce, requested harsher testing standards. In July, the last set of rules was approved.
Use the industry to provide a preliminary picture of what is feasible such that it is useful. But Smith cautioned, “don’t take our word as gospel.” Government officials should urge us to advance farther, in his opinion. At the White House, that’s what took place. Similar events, in his opinion, will occur in Congress and other international capitals.
