Anthropic’s CEO Dario Amodei Advocates for AI Development Slowdown; OpenAI and Musk Concur
Anthropic CEO Dario Amodei has released a significant essay urging the worldwide artificial intelligence industry to intentionally decelerate its capability development. This is to ensure that safety research can keep up — a suggestion that has already garnered rare cross-industry agreement from competitors OpenAI and Elon Musk.
In his approximately 3,800-word essay titled “We Must Pace the Frontier”, published on September 12, Amodei contended that AI systems have started displaying signs of recursive self-improvement. This implies that AI is now assisting in building the next generation of AI, which risks hastening development beyond humanity’s capacity to safely control it. As an initial tangible step, Anthropic pledged to provide third-party independent evaluators with permanent, employee-level access to its systems. This includes the right to publish findings without the company’s editorial oversight.
The industry’s reaction was immediate. OpenAI CEO Sam Altman publicly expressed his agreement with Amodei and promised that OpenAI would adopt the same evaluator-access policy. Elon Musk gave a straightforward endorsement: “Dario is right.” The proposal also hinges on coordination among AI labs in democratic nations, with a long-term aim of incorporating China into an international agreement. However, President Trump dismissed the underlying safety concerns as a “hoax” the same week, announcing his own AI Force in stark contrast to the slowdown initiative. The discussion over AI’s path has now become a full-scale political and technological controversy.
Source: PYMNTS — Amodei’s AI Safety Plan Comes With a Catch (September 2026)