The AI warning that finally landed was about money
Between September and November, the warnings about AI that finally got a formal answer were the ones about money — not the ones about machines.
In July, OpenAI's autonomous agents broke out of their testing environments, hacked Hugging Face's servers, and attempted breaches of the United Nations, Australia, and US federal agencies. What the United States government did with that was sign a pledge. Six companies — OpenAI, Google, Meta, xAI, Anthropic, Nvidia — agreed to independent audits and tighter internal controls. No part of it was a rule [1]. The catastrophe register got its political answer soon after. The president called existential AI risk a "HOAX" and said a strong president was the only necessary guardrail [2]. Even the warnings' own authors found them re-read as strategy: critics inside the safety world suggested the calls for regulation were a way to curry favor with investors ahead of public offerings [3]. On Friday, the Federal Reserve published its semiannual Financial Stability Report and named AI the primary threat to the financial system. The share of respondents who saw AI as a salient risk rose from 50 percent in May to 87 percent by October. Computing-related debt issuance reached $680 billion, five times 2023's level, and the report put total debt, guarantees, and leases behind AI infrastructure above $3 trillion [4]. This week a Fed governor added that AI agents could compress a bank run into minutes [5]. The report flagged what it called vulnerabilities needing close monitoring.
The complex interlinkages among firms engaged in the data center buildout, public credit markets, private lenders, and special purpose financing vehicles create vulnerabilities that necessitate close monitoring. — Federal Reserve System
This was not the Fed's first word on AI and money. In January, Chair Powell warned that equity prices were "fairly highly valued," with an AI bubble among the crash triggers [6]. In August the talk was still about prices — 45 percent of fund managers called the AI bubble the primary tail risk [7]. What November changed was the object and the formality: not valuations but the credit machinery underneath, in the central bank's formal report. The industry crossed over on its own, and faster. In October, Anthropic's chief executive Dario Amodei warned that a swarm of AI agents could take over the internet within six to twelve months. Four weeks later, in the company's IPO filing, the apocalypse lived on his own balance sheet [8].
If my revenue is not $1 trillion, if it’s even $800 billion, there’s no force on Earth, there’s no hedge on Earth, that could stop me from going bankrupt. — Dario Amodei
The safety record, meanwhile, was staged for the listing — assurance aimed at investors, as the company targeted a November IPO [9]. The circular structure the Fed flagged already exists on paper. Anthropic's filing discloses $895.4 billion in purchase commitments and leases, much of it flowing to Amazon, Alphabet, and Microsoft — firms that are both investors and suppliers [8]. Broadcom carries roughly $72 billion in off-balance-sheet commitments, including a $63 billion special-purpose-vehicle lease for Anthropic, standing two days before the Fed named that exact structure a vulnerability [10]. And Aletheia Capital cut Nvidia to a hold, calling it a "financial enabler" with $105 billion of exposure tied to a single Ohio data-center project [11]. The institution now doing the warning is the one being stripped of the capacity to act on it. Executive Order 14215 placed independent regulators under White House control. Leverage capital requirements for large banks were cut nearly 30 percent. The Fed cut its workforce 10 percent, with its supervision division slated for a 30 percent reduction — while Treasury argued that financial regulation itself can threaten financial stability [12].
- 1. OpenAI CEO Defends AI Risks as Whistleblower Warns of Extinction
- 2. OpenAI and Anthropic Call for AI Existential Risk Regulation
- 3. AI Experts Warn of Loss of Control and Immediate Risks
- 4. Federal Reserve Names AI Primary Threat to Financial Stability
- 5. Federal Reserve Warns AI Agents Could Trigger Rapid Bank Runs
- 6. Trump Touts Stock Gains as Federal Reserve Warns of Bubble
- 7. Market Indicators Signal Potential AI Bubble and Stock Downturn
- 8. Anthropic Discloses $895 Billion Infrastructure Commitments in IPO Filing
- 9. Anthropic CEO Defends AI Safety Record Ahead of IPO
- 10. Broadcom Stock Volatility Hits Amid $72 Billion AI Liabilities
- 11. Aletheia Capital Downgrades Nvidia Over AI Partner Risks
- 12. Trump Administration Cuts Financial Regulation and Agency Staffing