Every racing driver’s dream: you pull ahead at full throttle, then call for a speed limit for everyone behind you. That was my first thought when warnings from leading US AI companies made the news. Almost all the coverage focused on whether we might lose control of AI and how dangerous these systems could soon become. Of course, many of the risks are real, and we need to talk about them openly. But over the past few weeks, a lot of fear has been whipped up around AI. That’s why I think it’s worth asking where these warnings are coming from and what those issuing them stand to gain. Europe still has a lot of catching up to do in AI. So what does it mean for us when the companies out in front are the very ones calling for a slower pace and new rules?
Thankfully, that question is now part of the public debate too. On Lanz & Precht, the German writer and philosopher Richard David Precht was quite explicit about his unease when leading companies begin “acting as regulators as well”, seeking to “set the binding rules for everyone else”. Handelsblatt saw a certain “double standard”, while Welt went as far as asking: “AI horror as a strategy?”
Those who write the rules protect their lead
In September, OpenAI called for mandatory national safety requirements, shared standards and rules for when development needs to be slowed or stopped. Smaller developers and open models should not be held back by this, it says. Sounds fair. But how is that supposed to work when the same compliance requirements that barely trouble a large company eat up a start-up’s hard-won growth capital?
Dario Amodei makes the strategic interest explicit. A coordinated slowdown should give leading AI developers time, he writes, “without sacrificing commercial advantage or the United States’ lead in AI”. In other words, development should slow down without putting the companies’ commercial advantage or America’s AI lead at risk.
So we in Europe mustn’t be naive. Fear is never a good guide. Let’s look at this calmly: today’s market leaders built their international dominance in a world with almost no rules. And with the AI Act, the EU has already created a far more extensive safety framework than either the United States or China. That is already putting the brakes on companies here compared with their international competitors.
So now the big AI entrepreneurs are stoking fear while also wanting a hand in writing the rules for the next stage?
That combination should sound familiar, especially in Germany. In the 1980s, the German government pushed for European emissions standards. Leading German carmakers had already invested heavily in catalytic converter technology. For their less prepared competitors, Germany’s demands meant an expensive effort to catch up. This did not go unnoticed in Brussels: the European Commissioner for Industry, Karl-Heinz Narjes, saw the combination of catalytic converter requirements and German tax incentives as giving German industry an unjustified advantage.
Of course, my own interests as the founder of NEURA Robotics also shape my thinking on this. AI is fundamental to what we do, because it makes the difference between conventional automation and cognitive robotics that can act independently and anticipate what comes next.
In our products, perception, AI models and mechanics have to work together so that people are safe alongside them, even when things don’t go to plan. That safety has to be built into the product: into sensors and motors, into models and the rules that govern them. We see this as part of our own responsibility and our business interest. After all, only safe products will succeed and stay successful. No rulebook can take over a manufacturer’s responsibility.
Volvo showed decades ago how well a sense of responsibility and business interests can work together. In 1959, engineer Nils Bohlin developed the modern three-point seat belt, and Volvo made the patent freely available for use around the world. The company gained a reputation for safety that endures to this day, precisely because it charged no patent royalties and allowed other manufacturers and their customers to benefit.
I assume that the major AI companies in the United States also act responsibly, if only out of self-interest. So what interests me most is the strategic thinking behind their statements.
Europe must not regulate against its own interests
Let’s take a step back. The EU AI Act was adopted long before these latest American demands. Internationally, the EU’s market rules sit alongside a US policy aimed at accelerating infrastructure, development and deployment. An American executive order issued in June takes a voluntary approach to frontier models and rules out mandatory government approvals.
So there is already a clear imbalance in international competition. While their government picks up the pace at home, US market leaders are calling internationally for more rules and oversight.
Europe does, after all, have a reputation for turning warnings and potential risks into bureaucratic hurdles as quickly as possible. If we now get caught up in the AI fear campaign of the US market leaders, we will be pursuing an industrial policy against our own interests in the name of safety. Multinational tech companies can spread high testing and compliance costs, or even fines, across billions in revenue. A start-up pays for them out of its next funding round. That money is then missing from development and market entry. If there are too few approved testing bodies, the wait itself becomes another hurdle. And if international standards are defined around the architectures of the market leaders, other approaches can quickly look like risky deviations.
Ever since founding NEURA Robotics, I have called for a constructive public discussion about AI’s effects on our society and our safety. Unfortunately, even on social issues, the debate often drifts unchecked towards anxiety. We need to change that. Enthusiasm for safe technology has to become our driving force again! Europe can set safety objectives while allowing different technical routes to achieving them. Smaller providers must also be able to demonstrate compliance without difficulty, and the requirements must apply to existing systems as well as new ones. At the same time, we need more confidence in our own established safety culture and in our companies. Labs around the world are already developing technologies that regulators can’t even begin to write rules for, because they don’t yet know they exist.
We know the cost of dependence
What really does worry me is that too much regulation could unintentionally steer us into new dependencies. Since the 2000s, we have seen time and again how costly it is to hesitate over technology. Europe dramatically underestimated how digital platforms would change commerce and public life. Today, Germany’s Federal Cartel Office describes Amazon as a central gateway to online retail. Its rules and algorithms influence the success of countless businesses. In the case of YouTube, the European Commission is examining how recommendation systems can shape public debate and elections.
These companies understood what people wanted at the time, took major risks and earned their success. But Europe has also earned its place at the back of the field. We only really woke up once the markets had already been carved up. The rules that followed often hit young businesses harder than established corporations. Studies of the General Data Protection Regulation (GDPR) show a decline in investment in European technology companies and fewer new apps from Europe.
We must not repeat that mistake with artificial intelligence. Monopolies emerge when others hesitate or cannot get past the barriers to entry. Then, once again, someone else ends up deciding who gets access to models and infrastructure, what that access costs and which standards apply. Our vision for AI cannot be limited to curbing someone else’s market power a few years from now. We can put AI to work for our own purposes.
We can put AI to work for our own purposes. To do that, we in Europe need to remain able to act. With caution, but without fear.