US Federal Communications Commission (FCC) Chairman Brendan Carr announced that the agency will pursue the directive from the White House to review and possibly block state regulations related to artificial intelligence (AI). At an industry summit, he was quoted as saying that the FCC has authority under Section 253 of the Communications Act related to the FCC having authority to step in if a state or local law prohibits the deployment of modern infrastructure.

It’s been a while since I read 47 U.S. Code § 253, and it turns out that the language in that section is short and succinct:

In general, no State or local statute or regulation, or other State or local legal requirement, may prohibit or have the effect of prohibiting the ability of any entity to provide any interstate or intrastate telecommunications service.

Nothing in this section shall affect the ability of a State to impose, on a competitively neutral basis and consistent with section 254 of this title, requirements necessary to preserve and advance universal service, protect the public safety and welfare, ensure the continued quality of telecommunications service, and safeguard the rights of consumers.

If, after notice and an opportunity for public comment, the Commission determines that a State or local government has permitted or imposed any statute, regulation, or legal requirement that violates subsection (a) or (b), the Commission shall preempt the enforcement of such statute, regulation, or legal requirement to the extent necessary to correct such violation or inconsistency.

The FCC faces a regulatory conundrum if it tries to use Section 253 to regulate AI. It’s clear that Section 253 only applies to telecommunications services.

There are only two paths I can think of that would enable AI to be considered as a communications service. The obvious one is if Congress passes a law that says it is. However, I imagine the folks in the AI industry would fight hard to avoid being regulated like a telecommunications service.

The more roundabout way would be for the FCC to deem broadband to be a telecommunications service, and then somehow determine that AI comes along for the ride. That doesn’t seem likely since Chairman Carr had plans to immediately declare that broadband is not a telecommunications service when he became Chairman. He was saved from having to do that when, in early January, the U.S. Court of Appeals for the Sixth Circuit overturned the ruling from the previous FCC that put broadband under Title II regulation. That meant that the FCC reverted to the rules made by the Ajit Pai FCC that said that broadband is not a telecommunications service and can’t be regulated as such.

It’s hard to see a logical path to regulate AI as a telecommunications service if broadband is not telecommunications. I can’t think of a logical argument to enable that. It’s also not easy to claim that AI is communications. AI is mostly computer calculations done in a data center. AI prompts are initiated by users and are carried to a data center via broadband, and the results are carried back to the customers via broadband. But there is no way in a fiber network that AI bits are different from any other kinds of bits.

I’ve never heard any argument that the other kinds of content that ride broadband networks can be considered to be a communications service. If AI is somehow considered to be communications, then so are responses from Google search, a movie from Netflix, and practically everything else that traverses the Internet. Every company that transmits content will vehemently oppose the idea that what it sends over the Internet is a regulated communication service.

The FCC faces an additional challenge in trying to overturn State AI regulations. Several recent Court rulings have weakened the FCC’s authority. Loper Bright Enterprises v. Raimondo overturned the Chevron Doctrine, which brings into question the ability of the FCC to enact laws that were not specifically mandated by Congress. This year, McLaughlin Chiropractic v. McKesson Corp gave District Courts more leeway to disagree with rulings made by federal agencies like the FCC.

It’s clear that the FCC is going to tackle the issue, but it’s going to take some interesting regulatory and legal gymnastics to pull this off. Perhaps the FCC’s goal is to just tamp down State regulations before they are enacted, but a Google search tells me that every State either has passed or is considering regulations related to AI.