When Geoffrey Hinton, widely known as the “Godfather of AI,” speaks, the tech world listens. For decades, he pushed neural networks forward when most of the field dismissed them as a dead end. Now, after leaving Google and raising alarms about the speed and direction of artificial intelligence, he’s doing something few insiders dare: calling out the entire industry that he helped create.

Hinton’s message is straightforward but unsettling. AI is accelerating faster than society can adapt. The competition among major tech companies has become a race without guardrails, each breakthrough pushing us deeper into territory we barely understand. The risks he talks about aren’t science fiction; they’re the predictable consequences of deploying powerful learning systems at global scale without the institutional infrastructure needed to govern them.

He points out that no corporation or government has a full grip on what advanced AI systems are capable of today, let alone what they may be capable of in five years. He worries that models are becoming too powerful, too general, and too unpredictable. The alignment problem — making sure advanced AI systems behave in ways humans intend — remains unsolved. And yet the world continues deploying these systems in high-stakes environments: healthcare, finance, defense, education, and national security.

But here’s the part Hinton didn’t emphasize enough: the problem isn’t just the technology. The deeper issue is the structure of the global ecosystem building it.

The AI race isn’t happening in a vacuum. It’s happening inside a geopolitical contest, a corporate arms race, and an economic system designed to reward speed, not caution. Even if researchers agree on best practices, companies are pushed to break those practices the moment a competitor gains an advantage. Innovation outpaces regulation, regulation outpaces public understanding, and public understanding outpaces political will. This isn’t simply a technological problem — it’s a societal architecture problem.

Hinton is right that AI poses real risks, but the missing piece is the recognition that these risks are amplified by the incentives of the institutions deploying it. Tech companies are rewarded for releasing models that dazzle investors, not for slowing down to ensure long-term stability. National governments are rewarded for developing strategic AI capabilities before rival nations, not for building global treaties that restrict their use. Startups are rewarded for pushing boundaries, not for restraint. No amount of technical alignment work can compensate for misaligned incentives on a global scale.

Another point Hinton underestimates is the inevitability of decentralization. The industry is rapidly shifting away from a world where a handful of corporations control model development. Open-source models, community-driven research, and low-cost compute are making advanced AI available far beyond Silicon Valley. This democratization is powerful, but it also complicates the safety conversation. You cannot regulate an industry by only regulating a few companies when the capabilities are diffusing worldwide.

Hinton calls for caution, but we also need a coherent strategy — one that acknowledges the complexity of governing a technology that evolves faster than policy, faster than norms, and faster than global cooperation. His concerns about runaway AI systems are real, but the more pressing threat may be runaway incentives driving reckless deployment.

The Godfather of AI is sounding the alarm, and the industry should listen. But we must look beyond the technology itself. AI will not destabilize society on its own. What destabilizes society is the gap between the power of our tools and the maturity of the systems that wield them. That gap is widening. And unless the world addresses the incentives driving the AI race — not just the science behind it — even the most accurate warnings may come too late.