In his June 2026 essay “Policy on the AI Exponential,” Dario Amodei argues that AI is advancing at a lightning pace toward what he calls a “country of geniuses in a data center,” while government policy making moves dangerously slowly. Because undeniable risks—such as the cyber vulnerabilities demonstrated by Claude Mythos Preview—are now apparent, Amodei insists that policymakers must move beyond mere transparency and take proactive, urgent action.
To address this, the essay outlines five key policy areas that need immediate reimagining:
- Regulation and Public Safety: Amodei proposes treating frontier AI models like airplanes by establishing an FAA-like agency to enforce mandatory third-party testing. The government should have the authority to block the release of models that present unacceptable risks in areas like cybersecurity, biological weapons, and autonomous loss of control.
- Macroeconomics and Tax Policy: AI could trigger hyper-growth but also cause severe, enduring job displacement. To ensure shared prosperity, Amodei recommends pro-employment incentives, wage insurance, and long-term macroeconomic support like universal basic income, which could be funded by the immense economic growth or capital gains taxes.
- Accelerating AI’s Positive Impact: To prevent slow regulatory bodies from bottlenecking AI’s massive potential benefits, agencies like the FDA must adapt to an accelerated pace of innovation. Amodei suggests reforming biomedical approvals to accept AI-simulated clinical trials, improved toxicology predictions, and surrogate endpoints to speed up drug discovery.
- The State and Civil Liberties: Powerful AI risks upsetting the balance of power and enabling high-tech autocracy. To protect democratic freedoms, policies must ban the domestic use of fully autonomous weapons, close domestic surveillance loopholes involving data brokers, and guarantee citizens access to capable AI to defend themselves during legal or regulatory actions against the government.
- Securing Leadership by Democracies: Recognizing AI as the dominant source of future military and economic power, Amodei calls for a global coalition of democracies. This coalition would tightly manage the AI supply chain by sharing chips and infrastructure among allies while restricting them from adversaries, coordinating on mutual defense, and sharing AI’s economic benefits worldwide.
Amodei concludes that the current public concern over AI’s real risks creates a unique window of opportunity to enact common-sense, bipartisan legislation before the technology’s exponential curve advances much further.
If you want to explore additional information, go to Anthropic’s “Policy on the AI Exponential” page.