When AI’s Biggest Companies Ask for a Slowdown, I Worry About Small Firms Like Mine

I run a small firm. I usually root for other small firms, too.

I like seeing a small team figure something out before a much larger company does. I like watching people build something useful without first raising enough money to hire three departments. AI has made more of that possible, and I want to see where it goes.

I see it in commercial sales and marketing. A small company can research prospects, develop campaigns, draft outreach, and prepare proposals with AI without first staffing an entire sales and marketing operation. The same goes for federal contracting business development: finding opportunities, working through solicitations, and preparing a response become more manageable for a small team. That gives firms like mine a chance to compete for business we might otherwise lack the staff to pursue.

Then there is the everyday cost of getting a business agreement started. A small firm can use AI to draft simple terms or do a first review of a contract before calling a law firm. I want the option to understand the document, flag questions, and get a workable draft together without a $2,000 legal bill being the price of getting started. When I do need counsel, I can bring a more focused question.

Those are meaningful advantages when I am deciding where a small firm's time and money should go. They are part of what I want to protect in this debate.

So when I saw The Hill's report on Anthropic CEO Dario Amodei calling for slower AI development, my first reaction was skepticism.

The companies already at the frontier want everyone to move more slowly. I want to know what that means for the companies trying to catch them.

I keep hearing the same message

The message reaching me from executives, commentators, and talking heads increasingly sounds like this: be afraid of AI. Be very afraid. We need to slow down.

That is my reading of the public conversation. Amodei's actual proposal deserves a closer look. In his essay, he calls for outside evaluators inside frontier labs, coordination among democratic countries, and eventual international coordination. He explicitly distinguishes pacing from stopping model training. Anthropic is committing to the evaluator step itself.

I can see the value of independent people checking a lab's safety claims. I buy services from these companies. I would like more visibility into what I am buying.

What makes me uneasy is the move from inspecting a company's work to coordinating how quickly an entire industry can advance. Amodei favors regulation covering frontier labs and discusses government support for coordination. Other major AI leaders have publicly backed the direction of his proposal, as Axios reported.

Agreement among the biggest suppliers deserves scrutiny from everyone who depends on them. Their technical expertise matters. So does their commercial position.

I see a barrier that large companies can afford

My concern with AI regulation is straightforward: it can turn the ability to afford compliance into a prerequisite for the ability to innovate.

Imagine a rule that requires an expensive evaluation, extensive documentation, and a lengthy review before a new entrant can compete. An established company can assign staff, pay the bill, and keep serving its existing customers. A smaller challenger may spend its remaining runway getting permission to reach its first ones.

The requirement can be identical on paper and have completely different consequences.

That is the part I think gets lost when people say everyone will follow the same rules. Everyone does not start with the same cash, legal staff, customer base, or access to policymakers. A delay that a large company can absorb might end a small company's attempt altogether.

I don't need to assume anyone is lying about safety to question that outcome. A sincere safety proposal can still protect the companies already ahead. I want the cost of entering the market examined as closely as the promised reduction in risk.

There is a distinction here for my own business. Zenpo builds with AI; we are not training a frontier model. Amodei's proposed requirements target frontier developers. I am not claiming his essay would make every small consultancy install a team of auditors.

But the competitive structure of that market reaches us anyway. If fewer model companies can challenge the leaders, I have fewer alternatives when a supplier changes its prices, limits access, or takes its product in a direction that doesn't serve my clients. I have already written about why I want the freedom to switch AI providers.

I want the next small model company to have a chance. Its success could give my small company more choices.

I don't expect China to wait

I also struggle with the assumption that slowing down here buys everyone more time.

China's published direction points toward expansion. Its March 2026 government work report called for faster adoption of AI agents and broader commercial deployment through its AI Plus initiative. The accompanying draft five-year plan included further work on advanced AI capabilities. That is the direction described in China's own official account.

That doesn't prove what every Chinese lab will do next. It gives me very little reason to expect them to slow down because American executives think the pace is uncomfortable.

Amodei acknowledges this problem himself: slowing too far could let Chinese projects pull ahead, and any global agreement would need credible verification. His proposal addresses that risk. I remain unconvinced that we can manage the gap precisely enough to make a coordinated slowdown work as intended.

If we make it harder for new firms here to develop and compete while firms elsewhere keep improving, we may change where the next advances happen. I don't see how that automatically makes us safer.

I want room to build

I want companies to test their systems, secure them, and take responsibility when they fail. I want evidence behind safety claims, including claims that slowing development will make us safer.

For me, a credible proposal needs to identify the harm it addresses, explain how the requirement reduces that harm, and show that a new entrant has a practical way to meet it. If the only workable path assumes the resources of today's biggest labs, I see a market being closed to challengers.

I would rather see accessible testing tools, shared research on failures, and requirements tied to demonstrated risks. Those are things a small team can use to improve its work. An indefinite instruction to slow down gives that team very little to build toward.

I am rooting for the people who haven't made it yet. I want them in this conversation before the companies that have already made it decide what responsible competition is allowed to look like.

Four questions I keep coming back to

If a safety rule leaves only today's largest companies able to compete, how should we judge whether it worked?

I would want to look at both the harm it prevented and the alternatives it kept from reaching the market. How would we measure the cost of a useful product that never gets built, or a safer approach that never gets funded?

What evidence would make me accept a slowdown despite the cost to smaller firms?

I would need a specific danger, a credible reason that extra time would reduce it, and a clear condition for resuming progress. What evidence should then make us revise or remove the restriction if it fails to deliver?

If China keeps advancing while we slow down, where does the risk go?

I would want to know whether we had reduced the danger or moved more of the capability beyond our influence. What would we need to verify internationally before accepting that trade?

Who should help write the rules when the firms with the most expertise also benefit from barriers to entry?

I want their expertise at the table. I also want challengers, independent researchers, and small businesses that rely on their products there. What would give those groups real influence over the rules, especially the firms that do not exist yet?