Dario Amodei, co‑founder and CEO of Anthropic, published a policy essay confronting what he describes as a deepening mismatch between the breakneck pace of advanced artificial intelligence development and the slow pace of policymaking.
The central argument of the essay, titled "Policy on the AI Exponential," is that the technology's capability is scaling so rapidly that existing regulatory frameworks are too slow and too fragmented to meaningfully shape its safe deployment. "AI is progressing extremely fast — much faster than the policy process was built to handle," he wrote.
Amodei went on to emphasize that voluntary transparency measures, once sufficient when AI risks were theoretical, no longer provide meaningful protection. "Anthropic has long advocated for transparency requirements for frontier AI, because the risks weren't yet clear enough to regulate precisely. That is no longer sufficient."
He added that frontier AI models are approaching what he described as the "end of the exponential," a point where capability growth could rapidly outstrip human understanding and societal preparedness. "The most surprising thing has been the lack of public recognition of how close we are to the end of the exponential."
Upon this scenario, Amodei proposed a policy for mandatory third-party testing of high-risk AI systems. He called for audits that evaluate several factors, including "cyber, bio, and autonomy risks — with the power to block or revoke deployment of models that pose catastrophic risks." He compared the approach to aviation regulation, suggesting that, like aircraft, AI systems should only operate after passing rigorous certification.
Amodei also addressed the broader societal impact of rapid AI adoption. "Humanity needs to wake up ... [this essay is an attempt] to jolt people awake," he wrote, emphasizing the potential for job displacement, economic inequality, and social disruption. He proposed frameworks for workforce adaptation, training support, and policy measures to cushion communities from AI-driven disruption.
"Humanity is about to be handed almost unimaginable power, and it is deeply unclear whether [we] possess the maturity to wield it," Amodei added. The essay coincides with the rollout of Claude Mythos 5, Anthropic's latest model, which was described by the company as having the "strongest cybersecurity capabilities of any model in the world."
Alongside Mythos 5 was the launch of Fable 5, the publicly available, "safe" version of Mythos 5 with more "robust safeguards" designed to prevent users from obtaining information that could facilitate cyberattacks or assist with sensitive biological and chemical research.
Anthropic also said in a recent blog post "recursive self-improvement," where AI can improve itself without human involvement, could come sooner than expected.
The company noted that new data shows its frontier models have increased their speed of coding, debugging and research. That situation, it added, could form a feedback loop in which the tools create better models on their own.
"The big story here is what we see are indications that, contrary to some popular opinion, AI progress is going to speed up in coming years rather than stay the same, or diminish," Anthropic's Jack Clark told Axios last week.
"As organizations, and eventually probably as societies, we need to figure out the tools to validate and verify that the stuff being done by these AI systems is correct and is aligned with human intentions aligned with a thriving society," he added.
