Senator Mark Kelly wants AI companies to pay into a fund to offset AI’s negative impacts

The list of AI’s potential effects on society are growing by the week.

Skyrocketing demand for new energy generation needed to power massive data centers are raising utility bills, and generators are spewing methane into the atmosphere. Chatbots are encouraging self-harm to vulnerable users. AI leaders are warning of massive labor shifts as whole categories of jobs will be automated out of existence.

With the Trump administration signaling a regulatory free-for-all for AI companies, other than voluntary measures, it’s unclear how these negative impacts can be blunted if they get out of hand.

Senator Mark Kelly is proposing a novel plan to tackle some of these problems: making the AI companies foot at least a part of the bill.

In a new proposal, Kelly calls for the top frontier AI companies to come together and all pitch a pile of money into what he calls the “AI Horizon Fund.”

Kelly is looking at the massive valuations of the leading AI companies and the hundreds of billions they’re throwing at AI infrastructure projects as a resource to fund the plan. Kelly writes:

“It’s common sense to tap the enormous profits of the big companies developing and deploying AI so innovation thrives, opportunity is shared, and every community benefits. The fund would leverage multiple options for generating sustainable revenues from industry. With guidance from educators, workers, unions, experts, and industry, those funds would then be reinvested into programs that train workers for high-demand careers, including those deploying the AI innovations of tomorrow and the infrastructure that supports them.”

The outline is light on specifics, and is presented as “a starting point for conversation.” It’s not clear if companies would be legally compelled to contribute, such as through a tax on AI profits.

Many other large issues could complicate such an endeavor, including a sharply divided Congress, an administration that has pulled top tech executives close into its orbit, and companies that aren’t likely to want to pour money into a government fund.

Among the ideas in the plan to help balance out the impact of AI:

• Supporting workers displaced by AI through enhanced unemployment insurance benefits.

• “Upskilling” and “reskilling” of American workers, in the mold of trade apprenticeship models.

• Making data centers pay: “Frontier AI companies that place heavy demands on public infrastructure or environmental resources need to not only offset these impacts but strengthen the systems and infrastructure on which they depend.”

• Funding ongoing research into the harms of AI, including effects on mental health, fraudulent uses of AI to prey on vulnerable populations, and the effects of bias.