Anthropic launches new research institute to address rapidly accelerating AI risks

Artificial intelligence company Anthropic has announced the launch of the Anthropic Institute, a new research initiative designed to study the societal risks and opportunities created by increasingly powerful AI systems.

The institute will draw on internal research from Anthropic’s frontier AI development efforts and share insights with policymakers, researchers, and the public as the world approaches what the company believes will be a new era of highly capable AI.

According to Anthropic, the pace of AI progress has accelerated dramatically in recent years. The company noted that it took roughly two years to release its first commercial AI model, but only three more years to produce systems capable of identifying serious cybersecurity vulnerabilities, performing complex real world work tasks, and even helping accelerate the development of future AI systems.

Anthropic leaders believe this momentum could lead to even more dramatic advancements in the near future.

“AI development is accelerating,” said Anthropic CEO Dario Amodei in previous remarks about the company’s vision for advanced AI systems. He has argued that transformative AI, capable of dramatically improving science, economic growth, and human productivity, may arrive sooner than many experts anticipate.

Preparing for a world with powerful AI

The Anthropic Institute aims to explore a wide range of questions about how powerful AI could reshape society.

These include how automation could transform labor markets and economies, how AI systems might influence social and political structures, and how governments and institutions should govern increasingly capable technologies.

Researchers will also examine potential threats introduced by advanced AI systems, including cybersecurity risks and the possibility that AI could begin improving its own capabilities through recursive development.

Anthropic says addressing these issues early will be critical if society hopes to harness the benefits of advanced AI while minimizing potential harms.

A cross-disciplinary research hub

The institute will be led by Anthropic co-founder Jack Clark, who will serve as the company’s Head of Public Benefit.

The organization brings together several existing research groups inside Anthropic, including the Frontier Red Team, which stress tests the limits of AI systems, the Societal Impacts team studying how AI is used in real-world contexts, and the Economic Research group examining the technology’s effect on labor markets and economic systems.

Anthropic says the institute will also launch new research efforts, including forecasting future AI capabilities and analyzing how advanced AI systems might interact with legal systems.

New researchers join the initiative

Several prominent academics and researchers are joining the institute at its launch.

Among them is Matt Botvinick, a resident fellow at Yale Law School and former research leader at Google DeepMind, who will lead work focused on AI and the rule of law. Economist Anton Korinek, currently a professor at the University of Virginia, will head research examining how transformative AI could fundamentally reshape economic activity.

Meanwhile, Zoë Hitzig, previously involved in AI economics research at OpenAI, will help connect economic analysis with the training and development of AI models.

A window into frontier AI development

Anthropic says the new institute will provide a rare perspective into the development of cutting-edge AI systems. Because it operates within a company actively building frontier models, the institute will have access to technical insights that are typically unavailable to outside researchers.

The company says it plans to publish findings and engage with workers, communities, and industries likely to be affected by AI-driven economic and technological change.