Microsoft calls for shielding child safety ahead of the future U.S. federal AI law

Microsoft has publicly called for child safety to be excluded from the scope of the future federal artificial intelligence legislation in the United States —the legislation that, if passed, could replace or block existing state laws. The position was laid out by Suhail Khan, the company's Senior Director of External Affairs, during a panel on children's privacy aimed at congressional advisers and the private sector.

The nuance is relevant. Khan specified that the company supports the 'preemption language' —the primacy of federal rules over state ones— but argued that 'there must be exceptions for children's safety and child protections, including at the state level.' In other words: a common rule for almost everything, except when it comes to protecting minors, where states would retain room to legislate and enforce their own rules. The company had already distanced itself from Meta, Google and other big tech firms by supporting a 2024 version of the Kids Online Safety Act (KOSA), a proposal that stirred controversy in the sector over its implications for content moderation and platform liability.

Behind the episode lies a structural debate: the tug-of-war between unified federal regulation and the proliferation of state laws. States such as California, Colorado and Illinois have passed or are processing their own rules on high-risk AI, algorithmic transparency and data protection. The industry, in general, prefers a single federal law that simplifies compliance; the problem is that an overly homogeneous framework can sweep away local protections that work. Child safety introduces here a political and social variable that is hard to ignore.

Beyond the commercial interests —which exist and should not be lost sight of in any corporate positioning—, the proposal has a defensible regulatory logic: the principle of not using federal harmonization as an excuse to lower the floor of protection for the most vulnerable. Online child safety is, moreover, one of the few areas with bipartisan consensus in Congress, which gives this stance real chances of prospering. The details are worth following: the available material comes from a note behind a paywall and does not allow the panel's arguments to be expanded. But the framework taking shape —federalism with safeguard clauses for minors— may end up being one of the most debated pieces of U.S. AI regulation.