An AI executive’s dire warnings about the future are chilling – but his solution is worse than the problem

To be fair, public regulators haven’t exactly dazzled us with their speed or savvy. The EU’s AI Act is a good step, but not enough on its own, and the U.S. regulatory frameworks are fragmented and mostly reactive. The average congressional hearing on AI is a grim parade of buzzwords and tech executives politely nodding at lawmakers who don’t understand what they’re talking about. China, where DeepSeek is based, has been more aggressive in some areas, but it’s hard to argue that centralized authoritarian control is the better model for tech governance. Surveillance concerns and speech limitations don’t get easier to swallow just because they have a human signing the rules.

The current state of regulation is uneven, inconsistent, and often too slow. But that doesn’t mean the answer is to hand over the reins to the developers like they are benevolently neutral. They are not your friends or your representatives. They are certainly not suited to be physical and civilizational caretakers of humanity. They are commercial actors with products to sell and quarterly metrics to hit. When push comes to shove, they’ll sand down any ethical qualms until they fit neatly inside a slide deck.

You can’t mitigate harm when the very act of mitigation threatens your business model. If an AI-powered hiring system turns out to be discriminatory, fixing it costs money. If an automated content generator floods the web with low-quality sludge, turning it off affects revenue. There’s no incentive to do the right thing unless someone forces their hand, and by that point, it’s usually too late.

You may like

The tech industry has shown repeatedly that it’s not equipped to self-regulate in a way that prioritizes the public good over private gain. In fact, the mere idea that the architects of disruption should also be in charge of constructing what replaces the old order should terrify anyone who’s ever been on the wrong side of a platform’s algorithm.

None of this is to say that AI doesn’t have incredible potential for good or that demanding safeguards means you’re anti-technology. Despite confusion over the term, it’s worth remembering that the Luddites weren’t against technology either; they were anti-exploitation. Their protests weren’t about looms, but about factory owners who used those looms to undercut skilled labor and impose miserable working conditions.

Chen Deli is right to ring the alarm, but wrong about who should hold the bell. Whistleblowers don’t tend to emerge from boardrooms. We don’t yet have a coherent framework for what responsible AI governance looks like. We have pieces, but no connective tissue to make those ideas stick, and we lack the political courage to impose them on the people with the most power.

Still, I’m not entirely pessimistic. The frameworks we need could exist. They could be built by coalitions of governments, civil society, independent researchers, and yes, even some principled voices from within the tech world. But they’ll only come into being if enough people demand them.

If the next decade really does bring the kind of transformation Deli predicts, we’ll need more than corporate promises. We’ll need rules with teeth to preserve the safety and dignity of humanity without trying to make it a product for sale.