Why Anthropic CEO Dario Amodei wants AI regulated like aviation and pharma

Anthropic CEO Dario Amodei argues advanced AI systems should be regulated like aviation and pharmaceuticals, citing risks to public safety, national security, and economic inequality. His proposal includes mandatory testing, government oversight, and measures to address job displacement and authoritarian misuse of AI technology.
Anthropic CEO Dario Amodei has called for AI regulation modeled after aviation and pharmaceutical oversight, warning that frontier AI systems pose significant public safety and national security risks. In his essay *Policy on the AI Exponential*, Amodei argues that current AI advancements—such as Anthropic’s Claude Mythos Preview—demonstrate capabilities that could threaten financial systems, critical infrastructure, and cybersecurity. He notes that risks extend beyond cybersecurity, potentially including biological threats and autonomous AI systems, requiring stricter controls akin to nuclear materials if unchecked. Amodei proposes a regulatory framework similar to the U.S. Federal Aviation Administration (FAA), mandating testing for advanced AI models above a compute threshold. Key measures include risk assessments for cybersecurity, biosecurity, AI autonomy, and automated research and development, with government authority to block or delay unsafe models. He emphasizes the need for security protocols, red-teaming, and continuous threat monitoring, enforced by government agencies or accredited independent bodies. The essay also addresses economic challenges, warning that AI-driven automation could displace most cognitive jobs, creating unprecedented inequality. Amodei suggests tracking job displacement, implementing pro-employment incentives, and exploring long-term income support mechanisms like universal capital accounts. He argues that AI companies should bear the cost of increased energy demand from data centers to mitigate broader economic impacts. Additionally, Amodei highlights risks to democracy and civil liberties, stating that authoritarian governments could exploit powerful AI for mass surveillance and personal data inference. Existing civil liberties frameworks, he argues, are ill-equipped to address these threats, necessitating proactive safeguards. The proposal underscores the urgency of aligning AI development with regulatory oversight to prevent misuse while fostering innovation responsibly.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.