Why Congress shouldn't require licenses to build AI products and avoid Sam Altman's regulatory moat

Congress should not create Sam Altman’s regulatory moat for AI

This week, the Senate Judiciary committee hosted a hearing on AI, which included Open AI CEO Sam Altman, IBM Chief Privacy and Trust Officer Christina Montgomery, and New York University professor Gary Marcus.

Many bad ideas were proposed during the hearing, but one of the leading proposals getting attention from both members of Congress and even some of the panelists was requiring a license to build artificial intelligence products.

Adopting and implementing such a proposal would undermine innovation, hinder competition by erecting significant compliance costs, and allow China to further close the gap on U.S. leadership in AI. It should be soundly rejected.

Licensing for tech isn’t new

The proposal of licensing for tech is not new. Recently, during a separate Judiciary hearing, Senator Lindsey Graham floated a similar licensing proposal for social media platforms. While social media and AI are fundamentally different, trying to require a license for either would be an egregious mistake as it is a backhanded attempt to license speech.

It is ironic that the same senator who complains about social media censorship also pushes a proposal that can very realistically lead to more censorship online. When the government is granted the power of licensure over an industry, it gives them an incredible tool to make companies comply with the government’s preferred courses of action.

Would conservative AI search engine, TUSK, get a license for GIPPR, its ChatGPT equivalent? Given the government’s recent track record on trying to control the dissemination of information online, it is not unreasonable to think that such heterodox innovations might be punished.

Challenging the assumptions driving AI licensing proposals

Is AI really in a “wild west” phase?

A lot of the forces driving such proposals operate under the assumption that AI is in the wild west, that there are no rules governing them whatsoever. Therefore, something must be done. But there are already existing laws on the books and existing agencies that can enforce those rules against these companies.

Don’t take my word for it: Last month, the Department of Justice, the Federal Trade Commission, the Consumer Financial Protection Bureau, and the Equal Employment Opportunity Commission released a joint statement stating they would enforce their respective laws and regulations on companies that are operating in this space. Let our existing agencies enforce existing laws and regulations before jetting off and creating more unnecessary confusion by adding more rules to the already massive books.

Do we really need to “do something”?

Another force driving the conversation was the need to do something. Some senators were claiming that the U.S. was falling behind in regulations here, citing the European Union’s AI Acts path to passage as a good thing.

However, when you dig beneath the surface of that proposal, you cannot ignore the lasting negative impact it will have on the development of AI in the EU. The U.S. should not base its tech policy on the EU’s, as the EU has no leg to stand on.

As Benjamin Mueller of Information Technology and Innovation Foundation notes, a small fraction of the largest tech firms exist in the EU, and in the emerging field of AI, a similar outcome is already emerging and might only get worse with the EU passing the AI Act.  As Senator Jon Ossoff correctly noted during the hearing, the reviews of the EU’s GDPR have received “mixed results”, which is an understatement.

Regulations will only crowd out competition

Lastly, Congress must be wary of pursuing regulation due to the adverse effects regulation has on competition. When regulations create compliance costs, they erect a barrier discouraging entrepreneurs from entering the space. Increased consolidation — which players are driving in this industry — is the result.

It is easy for Sam Altman to advocate a licensing regime when OpenAI is already baked in as an incumbent in the industry with billions in support at its disposal. The small startup that might be better has no incentive to become larger than a certain arbitrary size in order to avoid getting the compliance costs that come with success under such regimes. That leaves the technology, and by extension, Americans, worse off.

The one point that the senators, panelists, and everyone including myself can agree on is that AI possesses a transformative power to disrupt and reshape our society. However, it would be a massive mistake to give in to the fear of this technology and pursue solutions that would hinder it — and the country’s ability to remain a leader in this critical emerging industry.

America needs to double down on what separates it from the rest of the world and empower AI to reach its fullest potential.