Anthropic Backers Fund AI Watchdog Groups Shaping Policy Debate

Overlapping Networks in AI Safety Advocacy

Two nonprofit organizations positioned as independent voices on artificial intelligence safety share significant financial and advisory connections to backers of Anthropic, the company behind the Claude AI assistant.

The Tarbell Center and Model Evaluation and Threat Research (METR) have emerged as influential players in debates over AI regulation. Both groups receive funding from Dustin Moskovitz, Meta's cofounder, and Jaan Tallinn, who cofounded Skype. AI researcher Paul Christiano also connects to METR's work. All three figures were instrumental in Anthropic's early growth, according to a Washington Examiner investigation.

Anthropic CEO Dario Amodei highlighted METR in a recent essay, describing the organization as a solution for AI safety evaluation. He announced that METR's team would serve as "third-party evaluators" for Anthropic's systems and urged other major AI companies to adopt similar oversight.

Financial Ties and Early Investments

Moskovitz and Tallinn both invested in Anthropic shortly after the company's 2021 founding. Christiano served as an initial trustee for Anthropic's Long-Term Benefit Trust in 2023 and founded the Alignment Research Center, which has directly supported Anthropic's technical work. He later left the trust role to join the UK government's AI Safety Institute, which receives backing from Anthropic.

METR's website indicates that nine AI developers have adopted its "Responsible Scaling Policies," frameworks designed to determine when advancing certain AI systems becomes too risky without additional safeguards. The group argues that independent third-party oversight is necessary because advanced AI systems might pursue objectives misaligned with human interests.

Moskovitz donated to Christiano's Alignment Research Center in 2022, though METR told the New York Post that donation was restricted from directly benefiting METR. Tallinn has contributed more than $150 million across hundreds of projects that include METR, Time magazine reported.

Journalism Fellowship Under Scrutiny

The Tarbell Center operates a fellowship program that places technology reporters at established news organizations to cover AI topics. The Washington Examiner previously reported on this program and disclosed that it had discussed placing a Tarbell fellow in its own newsroom, though no arrangement materialized.

The center's largest donors include Coefficient Giving, led by Moskovitz, and Tallinn's Survival and Flourishing Fund. Anthropic cofounder and policy chief Jack Clark appears as a speaker in the fellowship program.

A Tarbell spokesperson emphasized the organization's editorial independence, stating that neither Tarbell nor its funders determine what fellows cover or what conclusions they reach. The group also maintains it does not accept direct donations from AI companies or their employees.

METR President Chris Painter similarly stressed his organization's independence, noting it does not accept donations from frontier AI companies or their employees. He said METR aims to bring information from inside AI companies into the public domain.

Why it matters

As Congress debates AI regulation, the independence of voices shaping that debate becomes critical. Organizations presenting themselves as watchdogs while sharing financial backers with the companies they evaluate raise questions about potential conflicts of interest. The overlapping networks could influence which AI safety concerns receive attention and which regulatory approaches gain traction, potentially benefiting companies whose supporters fund the oversight mechanisms.

Details were first reported by the Washington Examiner.