Would Anthropic’s proposed Public Benefit Fund actually help workers disrupted by AI?

Anthropic, the Silicon Valley artificial-intelligence behemoth, announced last month that it would share a portion of its profits with the public as part of a larger mission to give back to society.

The rationale was simple, according to a post on the company’s website: As AI technology advances and the potential to generate enormous economic value grows, so does the prospect of instability for workers. Committing 5 per cent of the Claude maker’s shares to the Anthropic Public Benefit Fund, with additional contributions linked to the company’s growth in value, would be a way for it to provide support to people affected by AI, instead of boosting executive pay.

On the surface, the idea appears benevolent. But economists and policy experts who have studied the philanthropic efforts of large corporations caution that the model proposed by Anthropic could be self-serving, doing little to directly help labour segments that may be affected by AI.

Anthropic chief executive officer Dario Amodei has repeatedly warned that AI risks affecting white-collar jobs. The company’s own internal data shows that its hiring of junior technical workers dropped dramatically last year, as it amplified its use of Claude AI agents.

“The idea proposed by Anthropic is very conceptual, without a lot of detail,” said Rafael Gomez, director of the Centre for Industrial Relations and Human Resources at the University of Toronto. “It plays to audiences, and maybe it can allay fears. But really, it is pre-empting any kind of government regulation by defining the regulation as they want it.”

Anthropic’s IPO timing in doubt due to AI backlash, sources say

Federal AI Minister Evan Solomon has repeatedly warned that governments ceding control of AI regulation to frontier platforms would be an abdication of democratic responsibility. So far, however, Canada does not have a dedicated set of laws that place guardrails on AI technology. Similarly, the United States has no single comprehensive law akin to the European Union’s Artificial Intelligence Act, which defines risk parameters of AI technology based on its threat to human rights and safety.

This leaves AI companies with a first-mover advantage in suggesting or developing regulations, observed Barry Eidlin, an associate professor of sociology at McGill University whose research focuses on inequality and social movements.

“The vision projected by AI executives is that we are headed for a world of mass unemployment, where there will be an underclass of people that will need to be propped up with universal basic income, and society will then have to rely on the benevolence of these corporations,” he said. “So, to me, when I see things like Anthropic’s Public Benefit Fund, I think it’s a public-relations exercise aimed at maintaining their control over how regulation might be crafted.”

Anthropic did not respond to questions from The Globe and Mail about criticism of the Public Benefit Fund.

Anthropic in talks to lease massive data centre campus outside Edmonton, sources say

OpenAI, Anthropic’s rival, has also committed to investing in the public. In May, the OpenAI Foundation, the non-profit entity that owns the company’s for-profit arm, announced a US$250-million commitment in the form of grants and direct work placements to help workers “navigate the disruption caused by AI technology.” So far, the company has distributed US$50-million to various organizations, including groups dedicated to local journalism.

Both OpenAI and Anthropic are structured as public benefit corporations – business entities that are set up with a dual mandate to generate returns for investors and benefits for the public. (The Globe is part of two coalitions of news publishers suing Anthropic and OpenAI in Ontario court for unauthorized use of their copyrighted materials.)

“That means they have to have a social benefit,” said Rosalie Wyonch, an economist with the C.D. Howe Institute. “They have committed to it. So the fact that they have announced funds and set up foundations is, in a way, a move to give investors clarity about their commitment to that dual mandate.”

Ms. Wyonch also added that the companies could be signalling to governments, ahead of going public, that they recognize the impact of their technology and are determined to help, as a shield against heavy-handed regulation.

“If there are concerns by governments and the public that AI could impact society negatively, then companies want to proactively be able to say that they have invested money in coming up with solutions and will deal with it,” she said. “It lowers the probability of political action that would directly force them to do specific things.”

U.S. Senator Bernie Sanders, a harsh and vocal critic of the pace of development of AI technology, proposed legislation this past June that would impose a one-time, 50-per-cent tax on the equity of AI companies with revenue of more than US$200-million. The money would then be put in a sovereign wealth fund and make the American public a major shareholder in the country’s largest AI firms.

Prof. Gomez, the industrial relations expert, pointed to past instances of industry, labour and government working together to alleviate the impact of economic change on workers. The Canadian Skills Training and Employment Coalition, for example, began in 1986 as a joint venture between steel producers and the United Steelworkers union to collectively analyze what to do about Canadian steelworkers who were being displaced from their jobs because of technological change and offshoring to Asia.

“It took years of work to come up with public policy that satisfied all sides,” he said. “But it is possible to do. They just need to consult the right voices, across all spectrums.”