Header image: Renesas Silicon Valley Offices in the Edenvale district of San Jose, California 2012 by Cristiano Tomás, CC BY-SA 4.0, via Wikimedia Commons — cropped to 16:9 and colour-adjusted.
Key takeaways
- Over 1,000 AI lab employees signed a public letter demanding caution
- Incumbents benefit from slowing down while startups face survival threats
- US may use safety as a regulatory tool to set global AI standards
Over 1,000 AI lab employees put their names to a public letter in July. The brief doesn’t say. But scale like that doesn’t happen by accident. Something spooked them. Employee activism in AI labs is rare. When it hits this level, pay attention.
And the industry listened. CEOs of major AI companies are now publicly suggesting slower progress due to safety concerns. That’s new. For two decades, tech leadership has lived by one rule: move fast and break things. Now, suddenly, the rule is maybe don’t.
From ‘Move Fast’ to ‘Maybe Don’t’: The Mentality Shift
This isn’t just policy. It’s culture flipping upside down. The industry that once embraced ‘move fast and break things’ is now preaching caution. But is this shift organic? Or is it a script written by the winners?
The timing stinks. The AI race has been accelerating for years, safety be damned. Convenient. Startups and open-source labs can’t afford to slow down—they need breakthroughs to survive. Incumbents? They can wait. A safety pause doesn’t hurt them. It hurts their competitors.
The China Card: Why Geopolitics Made Slowing Down Unthinkable
For years, AI’s loudest voices waved the China flag. The argument was simple: if the US slows down, China won’t. And if China gets ahead, the consequences could be catastrophic. That argument won. It justified reckless speed. It silenced critics. It made caution look like surrender.
Now, suddenly, the industry is slowing down. The brief doesn’t say if the China argument has weakened. But the slowdown suggests it’s no longer decisive. Did the calculus shift? Or is this just a pivot that suits the incumbents?
If the China threat was real, why is the industry comfortable with a slowdown now? One possibility: the incumbents have already won. They don’t need to move fast anymore. They just need to keep everyone else from catching up.
The Cartel Question: Who Benefits from a Slower Pace?
Critics call it an anti-competitive cartel. The brief doesn’t name which companies are pushing hardest for caution. A pause doesn’t threaten them. It freezes the market in place.
Startups? They’re screwed. They need momentum to attract funding. They need breakthroughs to compete. A safety pause doesn’t just slow them down—it kills them. And if the pause becomes permanent, the incumbents win by default.
Is this ethics aligning with self-interest? Maybe. But the overlap is hard to ignore. When the biggest players call for a slowdown, ask: who benefits?
The Regulatory Angle: Sacha Haworth’s ‘Compass’ Pitch
Sacha Haworth of the Tech Oversight Project sees the slowdown as an opening. The US, she argues, can set the global standard for AI development. The brief doesn’t detail what that standard might look like. But the implication is clear: a more deliberate, safety-conscious approach.
Smart play. If the US slows down, it forces the rest of the world to follow. The US has a choice: compete on speed or compete on safety. Right now, it’s choosing safety.
But there’s risk. Slow down too much, and the US cedes leadership to China. The brief doesn’t say if that’s a real concern. But it’s worth considering. The China threat hasn’t disappeared. It’s just been reframed.
The Startup Dilemma: Caught Between Caution and Survival
The brief doesn’t mention startups directly. But they’re the ones who stand to lose the most. Smaller labs can’t afford long safety pauses. They need to move fast—funding, talent, competition. A pause doesn’t just slow them down. It starves them.
This creates a perverse incentive. The safer the industry becomes, the harder it is for startups to survive. Incumbents can play the long game. Startups can’t. If the slowdown holds, expect consolidation—fewer labs, fewer breakthroughs, fewer voices shaping AI’s future.
What’s Next: Safety vs. Speed as a Permanent Divide?
The brief doesn’t outline next steps. But the split between caution and acceleration could define the next phase of AI. Will the slowdown hold? Or will competition—or China—force a return to breakneck speed?
This isn’t just a tactical pause. It’s a fork in the road. The industry may never go back to move fast and break things. But if it doesn’t, who decides what replaces it? The incumbents? The regulators? The startups fighting for survival?
The most interesting question isn’t whether the slowdown is real. It’s who gets to control what comes next. And right now, the answer isn’t clear.