Microsoft’s Brad Smith pushes for licensing requirements for risky AI systems.
Microsoft wants to see federal licenses for powerful AI tools that could pose a public risk, the company’s president said on Politico Tech in an interview (episode link). The company is calling on Congress to impose new AI rules, Brad Smith said. He also thinks that others in the regulation-averse tech industry should get on board.
Endorsing the idea of a federal licensing regime as well as a new agency that can regulate the more powerful AI tools and platforms, Smith took a diversion from the typical corporate line that sees lower penalties and more freedom in developing as a good sign.
The US Congress is taking the threat and risks associated with artificial intelligence tools such as misinformation, disinformation, deepfakes, etc. very seriously with a series of hearings scheduled this week itself.
Four days after the interview, Smith joined leading AI personalities including Meta’s Mark Zuckerberg, OpenAI’s Sam Altman, Microsoft’s Satya Nadella, and Tesla’s Elon Musk in a closed-door session with the US Senate Majority Leader Chuck Schumer.
The meeting was held yesterday and Smith’s push to regulate AI like cars and drugs was likely to be on the agenda as he wanted Schumer to think about the growing concerns regarding the fast-paced development of AI tools, most notably generative AI platforms.
It’s also worth noting that last Tuesday, Smith testified in the Senate supporting a regulatory framework that was proposed by senators Richard Blumenthal and Josh Hawley to create a licensing entity for sophisticated or potentially dangerous AI models. The proposal also outlined that the licensing entity will hold companies accountable when their AI models and tools “breach privacy, violate civil rights, or otherwise cause cognizable harms.”
We’re moving toward a world of licensing and regulating AI capabilities as the world wakes up to the many dangers, risks, and threats posed by the mass proliferation and access to artificial intelligence and machine learning tools. However, it’s important to not overly regulate or license companies or tools that are doing good work, such as in science and medicine.
On that note, the Microsoft executive did say in the Politico interview that only risky AI systems should require licenses. He also mentioned that he expects both, Microsoft and OpenAI, to fall in this category and be subject to the licensing rules. He said that the company is “prepared to practice what we preach.”