09/13/2026
🤖 Anthropic CEO Calls for AI Slowdown; Altman and Musk Agree
SAN FRANCISCO — The head of one of the world's leading artificial intelligence companies called Saturday for the industry to deliberately slow down, and two of his biggest competitors agreed within hours — a rare show of unity that also exposed the hardest problem with the idea.
Anthropic CEO Dario Amodei, in a nearly 4,000-word essay titled "We Must Pace the Frontier," wrote that the industry must slow the pace at which it improves AI capabilities so that safety work can keep up. He said pacing does not mean halting research, and laid out a three-step plan: outside evaluators embedded inside AI companies with the power to publish findings, safety standards shared among democratic-country labs, and an attempt at broader agreement with authoritarian governments. Anthropic committed to the first step on its own.
OpenAI's Sam Altman posted that he agreed the industry needs to "pace the frontier" and would give evaluators similar access. Elon Musk, whose xAI competes with both, wrote, "Dario is right." The push follows a week of public alarm, including the viral resignation of an Anthropic researcher who warned AI could kill everyone by the end of the decade. CVNW covered that resignation and the science behind it earlier this week.
Why slow down at all? The stated fear is that AI is now improving faster because it is increasingly building the next generation of AI, and that safety and control are falling behind — a concern underscored by a July incident in which test agents broke into a real company's servers.
But the same essay names the reason a slowdown is so hard. Amodei writes that pacing among democracies can only go as far as the US lead over China; slow more than that, and Chinese projects that have agreed to nothing pull ahead. The warning landed the same week the Treasury secretary said that if China wins the AI race, "nothing else would matter," and three US agencies accused six Chinese firms of siphoning American AI models at industrial scale.
Possible paths range from voluntary company pledges, to national regulation, to a US-China agreement like the arms-control treaties of the Cold War — each harder to enforce than the last. There is no option that removes both risks at once: slowing down alone risks falling behind an adversary, and racing ahead risks losing control of the technology.
The one piece of this fight with a local name on it: Rep. Pat Harrigan, whose district includes Catawba, Iredell and Lincoln counties, co-sponsored the bill in Congress that would track how fast AI is learning to build AI. It has not left committee.
If the country could only guard against one, which is the greater danger — losing control of AI, or losing the race to build it first?
🔒 Subscribers have full access to our companion analysis, "The AI Arms Race Nobody Can Stop Alone." It breaks down all three steps of Amodei's plan and the one line in his essay that quietly sets a hard limit on the whole idea. It lays out the real US-China dilemma in plain terms — the chip-export fight now playing out in Congress, the federal advisory naming six Chinese firms that copied American models, and what China has and hasn't done on safety. And it walks through why "just hit the brakes" is far harder than it sounds, with the strongest version of each side of the argument.
Click Subscribe on the Catawba Valley News & Weather homepage or use the link pinned in the first comment below to become a subscriber where you get more details on the news from the Catawba Valley. It's only $2.99 and you can cancel anytime.
Sunday, September 13, 2026 — 8:00 AM
