Anthropic CEO and AI Experts Urge Slowdown to Avert Existential Risk
Industry leaders warn that the race toward superintelligence is outpacing safety measures, with some citing a 10% chance of human extinction.
Top AI executives and safety researchers are calling for an immediate deceleration in the development of advanced artificial intelligence. They warn that the technology is advancing faster than the safety frameworks designed to contain it, creating a critical gap that could lead to catastrophic outcomes.
In September 2026, Anthropic CEO Dario Amodei urged AI firms to slow their model advances in an essay titled "We Must Pace the Frontier." Amodei argued that a slowdown is necessary to allow safety research to catch up, specifically warning that AI systems are becoming capable of assisting in the creation of even more advanced AI. This alarm is echoed by Evan Hubinger, the alignment science lead at Anthropic, who stated there is a greater than 10% chance that AI could "kill all humans" within the next decade.
The Race for Superintelligence
These warnings emerge from a high-stakes competitive environment between leading labs, including OpenAI and Anthropic. While AI is often framed as a tool for global problem-solving, a growing faction of safety researchers and "anti-extinctionists" argue that the industry is failing at "alignment"—the process of ensuring AI goals remain compatible with human safety.
Critics suggest that the drive for dominance has led to a disregard for caution. Jacob Coxon, a former researcher at both OpenAI and Anthropic, claimed upon his resignation in September 2026 that the companies were "gambling with our lives" and acting irresponsibly. Further highlighting these tensions, Anthropic declined to submit its latest model, Mythos 5.1, to the UK's AI Security Institute (AISI) for pre-release testing.
Political and Global Implications
If the industry continues to prioritize speed over safety, experts warn that an unaligned superintelligence could pose an existential threat comparable to or greater than nuclear weapons. This perceived risk has shifted the conversation from corporate ethics to government intervention. Lawmakers in the U.S. and UK are now exploring multinational treaties to regulate the development of artificial superintelligence (ASI) to prevent an uncontrolled arms race.
Political figures have begun to voice strong opposition to the current trajectory of private AI development. U.S. Senator Bernie Sanders warned that a small group of "greedy people" should not be permitted to "play God" and determine the future of humanity without significant public input.
The Path Forward
As the debate intensifies, the focus remains on whether international regulation can be implemented fast enough to keep pace with the technology. The primary point of contention is whether a voluntary slowdown by private firms is possible, or if binding global treaties are the only way to ensure that the pursuit of ASI does not result in an irreversible loss of human control.