Remember When 1,100 AI Employees Asked for a Brake Pedal? Both Governments Said No.

Back in July we told you about a remarkable moment — more than 1,100 employees at OpenAI, Anthropic, Google DeepMind, and Meta signed an open letter asking the US government to help build tools that could pace AI development, essentially a brake pedal to use if capability ever outran anyone’s ability to safely manage it. This week that ask hit political reality, and both major governments in the global AI race rejected it. President Trump dismissed calls to pace frontier AI development and phoned Nvidia’s CEO to signal the opposite: full speed ahead. China separately rejected the same slowdown push. In a single day, the industry’s own request for a mechanism to slow down got shut down from both directions.

What Actually Happened This Week

The pacing letter we covered in July was notable specifically because it came from insiders at the companies with the most to lose commercially from slowing down. That made it a meaningful signal at the time. What happened this week tells you what that signal is actually worth in practice. When the request moved from an open letter into the realm of actual political and geopolitical decision-making, it ran into exactly the incentive structure you would expect. Neither government controlling the two largest concentrations of AI development capability has any interest in unilaterally slowing down while the other keeps racing. Trump’s phone call to Nvidia’s CEO was not a subtle signal — it was a direct statement that the US position is to keep accelerating.

This is not really a surprise once you think through the game theory. A pacing mechanism only works if it is coordinated — if one side pumps the brakes while the other does not, the side that slowed down simply falls behind. Both governments understanding this and rejecting unilateral pacing is the predictable outcome of that dynamic, not an unexpected plot twist.

Why This Matters More Than the Original Letter Did

The pacing letter in July was aspirational — a signal of concern from people close to the technology. This week’s rejection is the actual test of whether that concern translates into anything real. It does not, at least not yet, and not through the mechanism the letter proposed. If you have been operating under an assumption that AI development would eventually get externally paced by some coordinated government intervention — giving the market and your own planning time to catch up — this week is a clear signal that assumption does not hold in the current environment.

That does not mean no regulation is coming. We have covered plenty of regulatory activity this year — state-level data center rules, security disclosure requirements, various AI Act provisions. Regulation targeting specific harms, specific practices, specific disclosures continues to move forward in fits and starts. What this week rules out is something different: a deliberate, coordinated slowdown of the underlying pace of capability development itself. That specific mechanism just got rejected by the two entities capable of actually implementing it.

What This Means for How You Plan

If your business planning has any assumption baked in that AI capability development will naturally slow down and give the market time to stabilize, this week is worth revisiting that assumption directly. The signal here points the opposite direction — continued acceleration, not because anyone thinks that is risk-free, but because the competitive incentive to keep racing is stronger than the coordination needed to jointly slow down.

The practical implication is not urgency or panic. It is calibration. Plans that assume a breathing room period where AI capability plateaus while everyone adjusts look less realistic after this week than they did in July. Plans that assume continued rapid change — and build in genuine adaptability rather than betting on a pause — are better aligned with what governments on both sides of this race are actually signaling.

The Honest Takeaway

The people closest to this technology asked for a coordinated way to slow down if things moved too fast. The two governments most capable of making that happen looked at the actual incentives involved and said no — each for reasons specific to not wanting to fall behind the other. That is not a reason for alarm; it is simply useful information about the actual trajectory you are planning around. The pace of change you have experienced over the past year is a better predictor of the next year than any assumption that a deliberate slowdown is coming to give everyone room to adjust.

---

If you want to think through how to build adaptability into your own plans given this trajectory, this connects directly: Anthropic Just Modeled the Economy Through 2030. Here’s What It Means If You Employ Knowledge Workers.