Amid its ongoing tussle with the Pentagon, Anthropic has launched a new research initiative to examine the societal risks and economic consequences of increasingly powerful artificial intelligence systems.
The initiative, called the Anthropic Institute, will draw on the company’s internal research to provide analysis that policymakers, researchers and the public can use as AI capabilities advance.
Anthropic CEO Dario Amodei has repeatedly warned that transformative AI systems could emerge within the next few years. He has been vocal about AI’s safe use and the jobs that AI may impact.
The company, in its blog, said the effort reflects the pace at which generative AI is improving. In the five years since Anthropic was founded, its models have progressed from early experimental systems to tools capable of identifying severe cybersecurity vulnerabilities, carrying out real-world tasks, and accelerating further AI development.
Anthropic believes progress will accelerate even further. It argues that improvements in AI are compounding over time, meaning far more powerful systems could arrive sooner than many observers expect.
According to the company, such systems could bring major benefits in areas including scientific research, economic development and productivity. But they will also raise significant questions for governments and societies.
Among the issues the institute will study are how AI could reshape labour markets and economic activity, what new security risks it may introduce, and how societies should determine the values embedded in AI systems. The organisation also plans to examine scenarios in which AI systems begin to improve themselves, and how such technologies should be governed.
The institute will be led by Co-Founder Jack Clark, who will take on a new role as the company’s Head of Public Benefit.
Its staff will include machine learning engineers, economists and social scientists drawn from several of Anthropic’s existing research groups. These include the Frontier Red Team, which stress-tests advanced AI models to explore the limits of their capabilities, the Societal Impacts team studying real-world uses of AI, and an Economic Research group tracking effects on jobs and economic systems.
The institute will also launch new research projects, including efforts to forecast AI progress and study how powerful AI systems may interact with legal frameworks.
Anthropic said the centre will seek to share insights from building frontier AI systems while also engaging with external stakeholders, including workers and industries likely to face disruption.
Alongside the launch, the company said it was expanding its public policy operations as governments around the world begin shaping rules for artificial intelligence.
The policy team will focus on areas where Anthropic has set priorities, including model safety and transparency, infrastructure investment, export controls and democratic governance of AI.
The expanded policy effort will be led by Sarah Heck, who will serve as the company’s Head of Public Policy. Heck previously worked at Stripe and earlier held roles at the White House National Security Council.
Anthropic said it will open its first office in Washington, DC, later this year as part of a broader effort to expand its global policy engagement.
ALSO READ: Zoom Expands Enterprise Agentic AI Platform
