The people building AGI are now publicly worried about AGI, but they need you to keep funding it anyway.
The Summary
- An Anthropic researcher resigned last week, intensifying debate over AI development speed among tech leaders who've already committed hundreds of billions to the technology
- Executives are attempting to signal AI safety concerns while simultaneously reassuring investors and customers their massive AI bets remain sound
- The contradiction is simple: you can't pump the brakes and the accelerator at the same time, but that's exactly what's being attempted
The Signal
Tech CEOs have poured hundreds of billions into AI infrastructure, building data centers and training models at a pace that would have seemed reckless five years ago. Now some of those same leaders are publicly musing about the need to slow down. The timing creates an obvious problem: investors don't typically reward companies for pumping the brakes on their core growth strategy.
The resignation of Jacob Coxon from Anthropic crystallized this tension. Here's a researcher leaving one of the "safety-focused" AI labs, presumably over concerns about the very thing the company claims to prioritize. When your safety team starts walking out, you've got a messaging problem that goes beyond quarterly earnings calls.
"They're keen to telegraph their concerns about the technology's potential perils and have offered various remedies that include slowing down its development."
The dance tech leaders are now performing looks like this:
- Signal deep concern about AI risks to maintain credibility with regulators and the concerned public
- Propose self-governance frameworks that sound serious but lack enforcement teeth
- Continue building at full speed because competitors aren't slowing down and capital markets reward growth
This isn't entirely cynical. Many of these executives likely do worry about what they're building. The problem is structural. Public companies answer to shareholders. Private companies answer to VCs who need exits. Nobody answers to the future.
The Implication
Watch what they build, not what they say. If a company truly believes its AI development poses existential risks, it would materially change its roadmap, not just its talking points. Actual safety costs speed and money. Anything less is theater designed to keep capital flowing while appearing responsible.
For anyone building in this space, the lesson is simpler: the regulatory environment will tighten as this contradiction becomes more obvious. The gap between stated concerns and actual behavior can't hold. Plan accordingly.