4 Questions About the Anthropic Public Benefit Fund
Another major move in the rush to give away money made from AI came this week when Anthropic announced that it had set aside 5% of the shares in its public benefit corporation for a company it fully owns, called the Anthropic Public Benefit Fund. The fund is set up as a limited liability company, or LLC.
The news came almost a year after OpenAI changed its structure and gave the OpenAI Foundation a $130 billion ownership stake in the company. This made it one of the largest charitable organizations of its kind, bigger even than those associated with Gates and Lilly. Despite its name, the OpenAI Foundation is a public charity, not a private foundation subject to the IRS rule requiring it to pay out at least 5% of its assets each year. If Anthropic's shareholders approve the plan and the company goes public with a $2 trillion valuation, the shares held by the Anthropic Public Benefit Fund could be worth $100 billion. Anthropic's board could also add another 2% of the company's shares to the fund each time the company's value rises by 50% above a set starting point.
In its announcement on October 18, the company wrote, “When combined with today’s announcement, roughly 20% of Anthropic shares will be dedicated to the public good, plus any additional percentage that could be added if the company reaches specific valuation milestones.”
Anthropic could become a publicly traded company as early as November. But the Anthropic Public Benefit Fund is very unlikely to start distributing billions of dollars anytime soon.
Like the OpenAI Foundation, which aims to give out $1 billion in grants over one year but has so far distributed about $200 million in grants, Anthropic's fund will need time to grow its operations and meet its ambitious charitable goals. The fund will also decide how much information to share about where its money goes. LLCs are not required to publicly name the people or organizations that receive their money.
It is also important to remember that the fund's value will depend on how Anthropic's shares perform after the company goes public. I suspect most Anthropic investors will take a long-term view. But the company's share price could follow the short-term path of SpaceX, whose stock price has fallen by about one-third since reaching its peak after its public offering.
Anthropic said it would share more information about the fund's structure and leadership in the coming months. Until then, here are some questions about this latest example of how wealth created by AI could change the world of philanthropy, or charitable giving.
What will the fund's leadership structure look like? And what will its leaders focus on?
One reason well-known philanthropists such as Mark Zuckerberg and Priscilla Chan, Laura and John Arnold, Melinda French Gates, and Laurene Powell Jobs use LLCs to manage their giving is that these companies let them support charities, political activities, and for-profit businesses through one organization.
Anthropic is now using the same kind of structure. In its announcement, the company said the Anthropic Public Benefit Fund “will have increased flexibility and scale, enabling us to pursue our mission in new ways: For example, we might explore large donations to traditional nonprofits or direct distributions to specific populations.” The company also set a limit on this freedom. It promised that “none of the Anthropic equity held by the LLC will be used for lobbying or political giving, in the U.S. or abroad.”
Anthropic's decision to avoid political spending, at least for now, “is a sensible choice and may give the effort more credibility,” said Priya Shanker, executive director of the Stanford Center on Philanthropy and Civil Society, in an email to Inside Philanthropy. “I hope Anthropic continues to be as thoughtful in the governance and internal policies it establishes for the fund.”
Another major question is who will run the Anthropic Public Benefit Fund. One common criticism of the OpenAI Foundation is that its leadership structure creates built-in conflicts of interest. The nonprofit OpenAI Foundation and the for-profit OpenAI Group public benefit corporation have boards made up of almost the same people. It remains to be seen whether Anthropic's fund will have fully independent leaders whose main focus is the public good.
There are also questions about how the Anthropic Public Benefit Fund will actually give away its money. These questions are especially important because the fund could be so large.
“A liquidity event doesn’t automatically translate to social impact,” said Mike Wang, a partner at the philanthropic advisory firm Building Impact Partners. “Who decides where to give? What’s the theory of action here? Is this meant to be funding for innovation, growth, a safety net? Each of those calls for a different approach and strategy.”
As Wang explains, the fund's plan for creating change could greatly affect how much money reaches nonprofits through traditional grants. If its leaders decide, as Larry Ellison recently did, that philanthropy is not a good way to create change, they could focus mainly on for-profit businesses. The fund could then become a large operation that invests in businesses expected to create social benefits, while offering little money to organizations that apply for grants.
Shanker also wants to know whether independent experts and different points of view will help guide the fund's plans. She is interested in how it will balance responding to the immediate effects of AI with making longer-term investments in society. “There is a real opportunity here to build something that is not only well resourced,” she said, “but also offers a blueprint for how corporations can translate the value generated by their commercial operations into tangible and lasting public benefit.”
Will the fund choose to be as open and transparent as possible?
Is using an LLC really the best way for the new fund to reach its stated goal of “investing in the public benefit?”
For one thing, LLCs do not have to follow the IRS rule that requires private foundations to pay out at least 5% of their assets each year. This may not be a major problem in practice, since wealthy technology donors who give through LLCs often give more than 5% of their assets through their private foundations. I hope that once the Anthropic Public Benefit Fund has enough staff, it will give away at least 5% of its value each year, if not more.
The problem is that, unlike private foundations, LLCs do not have to publicly reveal who receives their money. One could argue that being open about how money is used is itself part of serving the public good.
Another question is which people and organizations will receive the money. If some of the fund's decision-makers follow the ideas of effective altruism, for example, money could go to public health and international aid organizations struggling after the breakup of USAID. But the meaning of “public good” becomes less clear when the money goes to for-profit companies whose main purpose is to make money for their shareholders.
“It’s not particularly uncommon for deep-pocketed donors to use a model [LLC] like this, but it becomes the donor’s choice to be transparent,” Wang said. “I would encourage Anthropic to commit to a minimum pace of giving and to public reporting, which I think is consistent with the spirit of their intentions.”
Anthropic seems to understand at least some of these concerns. In its recent announcement, the company said it would “report on our progress to date against our mission in an upcoming public benefit report.” We will have to wait and see whether the report accounts for every dollar the fund spends.
How will the fund deal with AI's effects on jobs?
The fund is almost certain to address what Anthropic calls the effects of AI on the job market. The company warned that AI “could cause instability for many workers.” It said it has “already committed $200 million to studying and addressing the issue, but given the pace of AI progress, bolder approaches may be needed. The fund will have versatility to pursue them.”
One of the most interesting things about the Anthropic Public Benefit Fund is that, because it is an LLC, it could use an idea supported by effective altruism and give money directly to individuals.
For years, technology funders have supported direct cash payments, such as universal basic income, because robots may eventually replace a large part of the workforce. The rapid growth of AI over the past few years has increased fears that many people could lose their jobs. But some technology leaders, including people whose companies are helping bring about these changes, have described this future in almost idealistic terms. In making the case for universal basic income in 2018, Elon Musk said that in a future with widespread automation, “people will have time to do other things, more complex things, more interesting things.” He also said people would have more free time.
This is the same Elon Musk who responded to Anthropic's announcement of its public benefit fund by announcing plans to give every child in America one share of SpaceX stock. It is not clear whether Musk will follow through. His record of carrying out ambitious promises to support good causes is mixed at best. Still, his announcement, along with Michael and Susan Dell's $6.25 billion contribution to “Trump Accounts,” suggests that direct payments may be becoming more popular among very wealthy technology donors.
“Cash support during a job change is a tool that can work quickly and directly,” Wang said. Shanker agreed, saying the fund could help society deal with economic disruption caused by AI by supporting “individuals and communities experiencing the immediate effects of technological change through direct financial assistance” and “experiments with guaranteed income.”
The fund could also use more traditional grantmaking through organizations that qualify as tax-exempt charities under section 501(c)(3) of the tax code. It could follow approaches already used by foundations, such as helping entry-level workers deal with AI's effects; forming partnerships that respond to employers' needs and creating paths into careers less likely to be affected by AI-driven changes; and supporting policies that put the needs of students, teachers, and low-paid workers first.
Of course, Anthropic differs from other grantmakers trying to help society adjust to the changes brought by AI. Along with OpenAI, Anthropic is actually building the technology that is driving those changes. This creates an obvious contradiction, but it also gives Anthropic access to its own research into how AI systems affect jobs and, as the company has warned, pose “existential risks to humanity.” “Anthropic likely has a better line of sight than most in the workforce space as to which jobs AI is likely to disrupt the most,” Wang said. “Sharing that data and building responsive interventions in collaboration with institutions like community colleges could be powerful.”
Shanker also recommends that the fund work with local governments, community foundations, and other organizations that connect funders with people who need support. These partnerships could help individuals and communities facing economic disruption caused by AI. “This would allow it to scale up its giving as needs emerge, rather than having to build the necessary infrastructure, relationships and knowledge from scratch,” she said.
But the Anthropic Public Benefit Fund itself is more or less starting from scratch. We do not yet know whether its leaders will take on the less glamorous, hands-on work of building lasting partnerships and giving grants to nonprofits working in workforce development. As Wang points out, this field “tends to be fragmented and lacking in capacity.” He also recognizes that even $100 billion cannot “replace the work we rely on public systems for, like unemployment insurance and community colleges.” He believes the fund's “best role is probably to catalyze, innovate, create proof points for what works, and push public systems to scale those solutions.”
Let's hope the fund tells the public where its money goes so we can judge its work for ourselves.