A Wake-Up Call in the Age of Exponential AI
On June 10, 2026, Dario Amodei, CEO of the artificial intelligence company Anthropic, released a comprehensive essay titled 'Policy on the AI Exponential'. In it, Amodei draws a vivid analogy to The Lord of the Rings: the hobbits trying to wake Treebeard the Ent to defend their forest. Just as the Ents move slowly while the forest is under threat, Amodei argues that AI is accelerating at breakneck speed while political and legislative processes lag dangerously behind.
The essay arrives at a critical moment. According to Amodei, in just four years, AI models have gone from writing a single line of coherent code to writing the majority of code at leading AI companies. Breakthroughs in biology, physics, mathematics, finance, law, and translation are equally striking. The so-called scaling laws of AI—which predict exponential growth in general cognitive capabilities with increased computational power—now have over a decade of empirical evidence behind them.
“If these scaling laws continue for just one or two more years, we are likely to get what I have called Powerful AI, or 'a country of geniuses in a data center,'” Amodei wrote.
Concrete Risks: Cybersecurity and Beyond
Amodei points to recent evidence that the power and dangers of AI are undeniable. The most striking example is Claude Mythos Preview, which demonstrated that frontier models pose very real cybersecurity risks, with potential to disrupt the financial sector, critical infrastructure, and national security. “Mythos Preview revolutionized the global cybersecurity landscape,” he states, adding that biological risks and AI autonomy risks could follow soon.
Five Key Policy Areas for an AI World
The essay outlines five areas where policy must be rethought:
- Regulation and Public Safety: Amodei proposes a regulatory model inspired by the U.S. Federal Aviation Administration (FAA). AI models above a certain compute threshold should undergo mandatory third-party testing in four areas: cybersecurity, biological weapons, loss of control of AI systems, and automated R&D that could accelerate these risks. The government should have the power to block or deter deployment of models with unacceptable risks.
- Macroeconomics and Fiscal Policy: Amodei warns that AI could bring extremely rapid economic growth but also massive and lasting job displacement. He proposes measuring and tracking displacement, pro-employment incentives (such as wage insurance, tax incentives to avoid layoffs, and retraining), and long-term macroeconomic support, including mechanisms like universal basic income funded by taxes on relevant companies or capital gains.
- Accelerating AI's Positive Impact: In areas like biomedicine, Amodei fears that regulatory systems (FDA, EMA) are not prepared for the flood of new products. He proposes reforms to accept AI simulations in clinical trials, pharmacodynamic and pharmacokinetic models, toxicology prediction, synthetic control arms, and more.
- The State and Civil Liberties: Amodei warns that AI could be the ultimate tool for autocracy. He proposes liability rules for autonomous weapons, a ban on domestic use of fully autonomous weapons, closing the loophole of mass data collection by companies, and ensuring citizens have access to AI advice at least as capable as what the government uses in adverse actions.
- Securing Democratic Leadership: Amodei urges the formation of a global coalition of democracies focused on building AI according to shared values, managing the chip and semiconductor supply chain, coordinating risk response, sharing benefits, mutual defense, and rejecting AI-driven repression.
Anthropic Backs Words with Legislative Proposals
Alongside the essay, Anthropic published a legislative proposal on frontier model testing and a policy framework for job displacement, with substantial financial backing. Amodei noted these are “first steps to show we are serious.”
A Window of Opportunity
Amodei remains optimistic: “The confluence of clear and present evidence of AI risks, a first taste of AI's potential for economic value creation and disruption, and a notable public rejection of unregulated approaches have created a situation where policymakers are unusually open to forward-looking action.”
The full essay is available on Anthropic's website. Amodei thanked Allan Dafoe, Mariano-Florentino Cuéllar, Richard Fontaine, Buddy Shah, Vas Narasimhan, Matt Yglesias, Nick Beckstead, Jason Matheny, Brad Carson, and many others for their feedback.
Source: Original essay by Dario Amodei