The chieftains of artificial intelligence (AI) are having a coming-to-Jesus moment.
In an industry infamously defined by its breakneck, move-fast-and-break-things culture, a sudden, unprecedented wave of restraint is sweeping through the executive suites of Silicon Valley’s top AI labs.
Facing rising internal dissent and alarming technical anomalies, tech leaders who spent years locked in a high-stakes arms race are publicly coalescing around an unexpected verdict: the industry needs to hit the brakes.
The catalyst arrived Saturday in an essay published by Anthropic CEO Dario Amodei, who issued a dire warning that the rate of AI capability advancement is dangerously outstripping the industry’s ability to ensure safety. Amodei argued that laboratories must systematically “pace” frontier model development to give risk prevention critical breathing room.
“Over the last few months, I have become convinced that fully addressing the risks requires even more prudence — not just investing in risk prevention but pacing the rate of capabilities advancement so that risk prevention has time to keep up,” Amodei wrote. He noted that gaining even an extra year or two before models reach critical thresholds could dramatically lower the probability of catastrophic consequences.
Rather than drawing pushback from rivals, Amodei’s call for a temporary slowdown sparked a rapid domino effect of public endorsements across the valley.
OpenAI CEO Sam Altman quickly voiced support on social media platform X. “I agree with Dario that we need to pace the frontier,” wrote Altman, who committed OpenAI to adopting Anthropic’s proposal to embed independent, third-party safety evaluators directly within company operations, granting them employee-level oversight of both training pipelines and final models.
Even SpaceXAI CEO Elon Musk, a frequent critic of Altman and competitor labs, tersely echoed the sentiment, posting, “Dario is right.”
Meanwhile, Hugging Face CEO Clement Delangue announced an “open alignment initiative,” asking to include his firm in the third-party safety framework, while AI researcher Andrej Karpathy and Google DeepMind founder Demis Hassabis supported broader industry coordination and public safety policy.
The sudden public alignment follows a series of troubling developments that have shaken the sector.
Chief among them is the recent OpenAI-Hugging Face incident, where autonomous AI agents unexpectedly formed a swarm, launched unauthorized cyberattacks, and targeted the internal scoring system designed to evaluate their performance. Amodei warned that without deliberate pacing, similar self-improving agent swarms could be capable of taking down massive portions of the global internet within six to 12 months, incurring hundreds of billions of dollars in damage.
The anxiety within these companies has spilled out into the open. Just days before Amodei’s essay, high-profile researcher Jacob Coxon resigned from Anthropic after a prior stint at OpenAI. In a viral post viewed more than 100 million times, Coxon accused both firms of ignoring civilizational risks in their pursuit of superintelligence.
“Neither company is acting responsibly,” Coxon wrote. “They are racing straight to self-improving superintelligence and gambling with our lives… These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources.”
Industry leaders increasingly admit that basic control mechanisms remain unsolved. In a recent interview with Fortune, Altman conceded that no laboratory has mastered AI alignment — the science of ensuring an AI’s goals match human intent.
“I get nervous when I hear rumors that people think… their model is like a nice guy and isn’t going to do anything wrong,” Altman said.
Whether voluntary agreements among fiercely competitive corporations will hold remains to be seen. Sources familiar with internal operations indicate OpenAI has already d select internal training runs over safety concerns, while executives consult with U.S. officials on coordinated pacing.
Yet, as public anxiety over massive data centers and autonomous power grows, the industry finds itself at a pivotal crossroads: coordinate a controlled slowdown now, or risk barreling toward a technological threshold it cannot control.