AI is creating a new wave of philanthropists. The system they're walking into is broken

A new class of very rich AI builders is about to test a philanthropy system that already isn’t working very well. The people advising them say many of these future donors genuinely want to give well. The problem is that the machinery around big gifts nudges them toward delay, not action.

The familiar path is the donor-advised fund, or DAF. Put money in, claim the tax deduction, and decide on the recipient later. That can be later next month or later never. Once the deduction is secured, the pressure to make a real grant fades fast.

The numbers explain why critics are so uneasy. More than $300 billion in philanthropic capital is sitting in American DAF accounts, and only about a quarter of those assets is paid out in a typical year. Some of that movement is just money shifting from one DAF to another, which helps no beneficiaries at all. In 2024, Fidelity Charitable took in nearly $16 billion in contributions, making it the most successful charitable fundraiser in the United States. Eleven of the country’s top twenty fundraising “charities” are DAF sponsors.

This is not just a bad look; it’s a design problem. DAF sponsors generally earn fees based on assets under management, so they are rewarded when money stays parked. Fidelity has made more than $1 billion in revenue from its charitable arm over the last five years. Private foundations face a different rule and must give away at least 5 percent of their assets each year. DAFs have no equivalent requirement.

The source piece argues that the system’s original bargain was simple: society gives up tax revenue now, and charities get the money later. That bargain only works if the money actually reaches charities. With a wave of AI fortunes likely on the way, the risk is that the broken incentives just scale up along with the wealth.