Zuckerberg Says AI Labs Can Set Their Own Safety Pace
Anthropic Chief Executive Dario Amodei called on Sept. 12, 2026, for a coordinated effort to pace frontier AI development, warning that rapidly advancing capabilities and recursive self-improvement could outstrip safeguards. The debate has divided technology leaders over whether laboratories should slow collectively or manage risks independently, while raising broader questions about corporate accountability, outside oversight and the competitive pressures shaping increasingly autonomous AI systems.
Meta Chief Executive Mark Zuckerberg said on Sept. 15 that each laboratory has both the responsibility and incentive to adjust its own development pace when safety demands it. He cited Meta’s decision to delay the release of its Muse AI agent by several months for security work. Zuckerberg backed independent evaluations and argued that alignment and user trust will be competitive advantages, while urging developers to direct most computing capacity toward serving users instead of racing for recursive self-improvement.
All Coverage
2 original reportsThe Backstory
The history behind this eventAnthropic CEO Urges Slower Frontier AI Push
Frontier AI competition is shifting from chatbots toward agents that can act autonomously, write software and help design successor systems, raising concern that capabilities are outrunning alignment, testing and regulation. Anthropic Chief Executive Dario Amodei argues that “pacing” is not a halt to model training or technical progress, but a coordinated effort to give developers and independent reviewers more time to build safeguards. The challenge is a prisoner’s dilemma: any company or country slowing alone risks surrendering commercial and strategic ground.
In a September 12, 2026 essay titled “We Must Pace the Frontier,” Amodei warned that a misaligned agent swarm could seize much of the internet within six to 12 months and cause hundreds of billions of dollars in damage. His three-part plan calls for embedded third-party evaluators, coordination among AI labs in democracies, and eventual engagement with China and other authoritarian governments. Anthropic pledged to adopt the first measure immediately; OpenAI CEO Sam Altman said his company would follow, while Elon Musk and Google DeepMind CEO Demis Hassabis backed the broader proposal.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.
If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →