Four US senators want public companies to open their books on artificial intelligence. Their new bill would force disclosure of AI governance, financing deals and risks, after trillions of dollars flowed into the technology with little visibility into who bears the cost.
What the bill demands
The AUDIT AI Act, introduced by Senators Warner, Kennedy, Alsobrooks and Rounds, targets how publicly traded companies invest in, deploy and manage AI. It would direct the Securities and Exchange Commission to write new rules covering oversight, financing and risk.
Public companies would have to say whether they have an AI governance policy and a dedicated board committee, and explain why if they do not. Any policy they have would have to be published, so investors can judge how seriously a company oversees the technology.
The bill also reaches into the financing. Non-financial public companies would have to disclose AI-related joint ventures, special purpose vehicles, variable interest entities and other major debt arrangements in SEC filings, so the true scale of their AI spending stops hiding off the balance sheet.
Standard risks, in standard form
Companies with significant AI exposure would face standardised risk disclosure in annual reports, covering dependence on outside AI models, concentration among computing providers, changing costs and liabilities tied to data rights. Investors could then compare the same risks across companies.
The bill also asks the SEC to study how companies account for AI research, revenues and capital spending, and to report each year on how automated trading strategies affect prices, market efficiency and herd behaviour among investors. The findings would feed future policymaking.
Why now
The timing is the point. Brookings estimates that building AI infrastructure could require $10.3 trillion from 2025 to 2032, about 3.6 percent of US GDP each year. The senators argue that a slowdown in that buildout could shake the wider economy.
Warner warned that huge exposure could be building behind the scenes, discovered only when something goes wrong. He pointed to circular deals in which AI companies finance businesses that are also their customers, naming chipmaker Nvidia’s support for firms that buy its chips.
Our Read
This bill asks the bubble question nobody can answer yet. Trillions are flowing into AI, much of it through structures that keep debt and commitments off the balance sheet. Anthropic raised $81 billion in a single week; the senators want to know what the rest of the wave looks like.
Why It Matters
Disclosure does not stop a bubble, but it changes who is surprised by one. Standardised reporting would let investors compare AI exposure across companies instead of guessing from press releases. That matters most when the financing is as large and as tangled as the senators describe.
The bill is only a proposal, and it imposes nothing unless Congress passes it and the SEC writes the rules. But four senators from both parties, on the Banking Committee, have now put the AI financing question on the legislative record. The industry’s answer will be watched.
Background
Warner has pressed the issue for months, warning that insufficient transparency makes it hard to judge how widely an AI setback could spread. Former SEC officials backed the bill, including ex-commissioner Robert Jackson and former chief accountant Paul Munter, who called better disclosure essential.
Source: Senate press release


0 Comments