What happened
Anthropic CEO Dario Amodei published 'We Must Pace the Frontier' on September 12, 2026, proposing a three-step plan to slow AI development amid growing safety concerns about recursive self-improvement potentially arriving by early 2027.
Dario Amodei, CEO of Anthropic, published a long essay titled 'We Must Pace the Frontier' on Saturday, September 12, 2026, launching what became the most high-profile debate yet over whether AI development should be slowed down.
Amodei's three-step plan calls for embedding third-party evaluators inside AI labs, coordinating between frontier AI companies in democratic countries, and pursuing global coordination between democratic and authoritarian governments. He warned that recursive self-improvement — AI systems training and creating new versions of themselves — could arrive as early as early 2027.
OpenAI CEO Sam Altman endorsed the essay, while Google DeepMind co-founder Demis Hassabis and Elon Musk also backed the proposal. China's Foreign Ministry rejected the calls as a US Cold War playbook, President Trump opposed AI guardrails on Truth Social, and Meta CEO Mark Zuckerberg publicly rejected any coordinated slowdown.
Key Facts
Key facts about the AI slowdown debate.
- Essay published Sept 12, 2026
- RSI expected by early 2027
- 1,000+ employees signed letter
- Coxon letter: 170M+ views
What's Next
The coming months will test whether the pace-setting commitments are substantive or symbolic.
Why it matters for me
The debate has divided tech leaders, governments, and researchers — some endorsing the slowdown as essential for safety, others rejecting it as a cynical power play or unnecessary hindrance.
Safety · Work
Ai Ethics Researchers · Tech Industry
If recursive self-improvement arrives by early 2027 as Anthropic warns, it poses direct existential risk to humanity and reshapes the AI industry's trajectory.
Money
Tech Industry · General Public
Slowdown commitments could delay AI product launches and affect competitive positioning, but may also lead to new federal regulations.
Daily Life · Safety
General Public
The outcome of this debate will determine how AI systems are deployed in everyday life — from healthcare to transportation to education.
What to remember
Whether the AI industry actually slows down depends on whether coordination translates into enforceable commitments — and whether governments are willing to act.
AI leaders are debating whether to slow down development — here's what's at stake.
Verified sources (5)
The AI Superintelligence Slowdown (rolling tracker)
↗We Must Pace the Frontier
↗Is Big Tech's AI slowdown a safety pact or a cartel?
↗China rejects AI slowdown calls, blasts US tech leaders' 'fearmongering'
↗What execs and politicians are saying about slowing down AI development
↗Claims and linked sources
8 claimsAnthropic CEO Dario Amodei published 'We Must Pace the Frontier' on September 12, 2026, proposing a coordinated slowdown of AI development.
Amodei's three-step plan includes embedding third-party evaluators, coordinating among democracies, and pursuing global coordination.
OpenAI CEO Sam Altman endorsed Amodei's essay and committed to independent evaluators at OpenAI.
Anthropic researcher Jacob Coxon resigned with a letter warning of racing towards superintelligence; the letter received over 170 million views on X.
Over 1,000 AI lab employees signed a July 2026 letter calling for a slowdown after the OpenAI-Hugging Face rogue agent incident.
China rejected the slowdown calls, framing them as a US Cold War playbook; President Trump opposed AI guardrails on Truth Social.
Mark Zuckerberg rejected coordinated slowdown, arguing slowing releases risks American leadership.
Microsoft published a 37-page Humanist AI Code of Conduct stating people matter more than AI and models are not conscious.
Community signals
Signal-only · not reporting“Left unchecked, recursive self-improvement could outrun our ability to understand and control these systems.”
“I agree with Dario that we need to pace the frontier.”
“The only control or guardrails that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT.”
“We are racing straight to self-improving superintelligence and gambling with our lives.”