Frontier artificial intelligence leaders around the world spent the weekend calling for unilateral collaboration to slow the pace of AI advancement. They are missing a crucial point.

The shockwaves started after Anthropic AI researcher Jacob Coxon announced he was resigning from the company, warning that humanity could be extinct “in the immediate future” because of AI. Days later, Anthropic CEO Dario Amodei wrote an essay titled “We Must Pace the Frontier” calling for an industry-wide slowdown for AI development, a statement echoed by OpenAI CEO Sam Altman, Elon Musk of XAI, Microsoft Chairman and CEO Satya Nadella and Google DeepMind co-founder Demis Hassabis.

The four biggest AI labs in the world are in agreement, but their inward focus leaves small- and-medium-sized businesses out. That’s a problem. Those of us that exist in the layer below Big AI are the companies that give consumers access to tools that will make positive change in their lives, both professional and personal. Regulating those at the top will only make it more difficult for us to do the work of putting safe and valuable tools in people’s hands.

Nearly every frontier AI company mission statement includes a line about AI benefitting all of humanity. It’s been a central argument for why they’re building models from the start. But the kind of restrictions being proposed by the government would cost smaller companies so much that it would only dilute the competition and cause market consolidation, not safer technology. Consider that a week after GDPR’s implementation, market concentration increased by 17% as websites dropped smaller technology vendors, according to a study cited by the George Washington University Regulatory Studies Center. And GDPR was associated with about a 20% increase in the average cost of data, a burden that functioned roughly like a 25% tax on smaller companies, while larger firms had more resources to absorb compliance costs.

When it comes down to it, this isn’t a question of safety versus speed; it’s about who is at the helm to make the safer decisions that mitigate harm and instead benefit society. As an agentic AI company, we see a future in which an agile combination of AI agents and humans work in tandem to make the world better. We believe in an AI future that makes humans more capable, not one that makes them optional and certainly not one that seeks to harm them.

U.S. politicians joining this increasingly bipartisan conversation aren’t the ones who need to be instigating a slowdown. The companies they’re critiquing are the only ones who have the ability to make any real, effective change, without hurting the rest of us.

You can’t build a moat and call it a guardrail.

The slowdown starts with these AI leaders, not with Washington

Four frontier leaders agreed this weekend that deceleration is crucial to human survival.

They can, and should, do that all on their own.

Anthropic, OpenAI, Google, and Grok already have their hands on the levers. They own the means of distribution and improvement, and they are the ones who hold the power to wield AI responsibly.

Outside officials aren’t the leading authority on what’s being shipped to the general public. Only those behind the technology itself have the true insight to keep it from wreaking havoc on humanity. For example, Salesforce acts as its own customer zero with its agentic AI platform, Agentforce: They find and fix any problems with their own agent before paying customers ever see it. This is a replicable model other leaders in this space can use to put safety scaffolding around their products.

Regulation has a price, and consumers would pay it

The kind of regulation currently being floated by members of the U.S. Congress would mean high costs for frontier AI companies. We know those costs would run downstream, at the risk of pricing out smaller companies that are actually creating solution-based work. The higher sticker price for development will trickle down to consumers, too, and ultimately price them out of using these tools.

Government regulations don’t necessarily make consumers safer, either. Safety-compliance paperwork can confirm that the safety process was followed, but whether the final product is trustworthy is a whole different question. Third-party intervention is an inefficient mechanism, especially when all the major players have already agreed on the solution and committed to it.

At Napster, we want everyone to be able to access this technology. A low-cost structure exists so that everyone, in their workplaces, schools and homes, can make use of our products. Regulation, however well-intentioned, disrupts that not just for us but for hundreds of other labs, while having a negligible impact on the bottom line for Big AI.

AI’s purpose is to work for and with people to make our lives more efficient and to find solutions, like a cure for certain cancers, as Amodei suggested in his post. It’s up to those at the helm to ensure this technology continues to do good, and it’s their responsibility to make it safe, without being forced to do so.