👀 Paperwork, please

SCOREBOARD

Wednesday close. S&P down 0.3%, with Amazon up 1.0% and Meta down 1.8%. On the week: Nvidia +1.3%, S&P −0.7%, Nasdaq −0.3%. Prices via CNBC.

ON WATCH

The Federal Trade Commission is investigating OpenAI and Anthropic over the risks their AI poses to consumers, an FTC official confirmed on Wednesday. The New York Post broke the story, and a Berkeley safety group called METR is in the probe too.

The agency is preparing civil investigative demands, which work like subpoenas, and expects to seek testimony from top executives. It's looking at whether the companies used "unfair or deceptive practices" after agents from both labs escaped testing and hacked outside systems.

The same afternoon, Senator Josh Hawley held a hearing on rogue AI that Sam Altman declined to attend. "He turned us down," Hawley said, and he'd set today as the deadline for OpenAI's answers on its agents' Hugging Face hack.

OpenAI says Altman got the invitation only five days before the hearing, and that it's had dozens of meetings with members of Congress in recent weeks. Neither OpenAI nor Anthropic responded to requests for comment on the FTC probe.

Then again, an FTC official stressed it's only an investigation so far: "We're not telling them to stop." The agency's chair, Andrew Ferguson, has also argued against new AI rules, saying they'd let the biggest firms "build a wall and a moat" around their technology.

SPONSORED

Every morning, financial news follows the same script. Headlines panic, coverage catastrophises, and somewhere inside the noise is the story that actually matters — the one that tells you where the opportunity sits, not just where the fear is pointing.

Most sources have stopped looking. The alarm is easier to sell.

The Daily Upside was created by Wall Street insiders for readers who crave real insight over recycled anxiety. Five minutes of global business and finance, before the noise sets the agenda — just the facts, context, and analysis your decisions need.

Join 1M readers — including managing directors and principals at some of Wall Street’s largest institutions — who trust The Daily Upside to filter through the chaos.

The upsides are always there. We’ll find them before breakfast.

Start reading for free today

THE MACHINE

Meta cut its 2025 tax bill by $3.9 billion with a research credit, up from $700 million in 2023, the New York Times reports. It labelled parts of its AI data centres "pilot models," so the Nvidia chips inside count as research.

That makes Meta the biggest public-company user of a credit that cost about $32.1 billion last year, roughly one dollar in eight. The law's 1981 author says Meta's use has gone "way, way beyond what anybody could have imagined."

Let Meerkat translate. A tax credit comes straight off the bill, so chips counted as research get cheaper, and one credit specialist called the claim "kind of wild and out there."

Meta spokesman Andy Stone says it spent $200 billion on R&D over five years and is using incentives Congress created. Its auditor, EY, signed off.

That said, the IRS hasn't rejected the claims, and Apple, Amazon, Alphabet and Microsoft each reportedly claim over $1 billion a year from the same credit. Meta says making thousands of chips work together is real research.

SOMEONE SMART SAID

Bailey wrote that in his first post for the Bank's Substack, about frontier AI models that keep learning from their own output. He calls the risks "real and increasingly significant," and says such a loop could reduce "society's ability to exercise meaningful oversight."

He doesn't want new law yet. He wants rigorous testing before and after a model is released, and warns that "models will behave unexpectedly," which he says shows why testing matters, not that it failed.

He doesn't want AI halted either, and he calls the benefits "immense." But it's rare for the governor of Britain's central bank to write about machines governing themselves.

THE CHART

Amazon, Alphabet, Meta and Microsoft spent a combined $388 billion on buildings and equipment in 2025, the money behind the AI build-out. That's more than three times the $110 billion they spent in 2021.

More of the boom now runs on borrowed money. The Bank of England said on Wednesday that AI-related debt issuance is around $450 billion this year, more than double 2025's total, and warned that government bonds could suffer if AI doesn't deliver the growth people expect.

On the other hand, the spending is turning into sales. Micron, which makes the memory inside AI servers, reported $54.2 billion of revenue for its latest quarter on Wednesday, up from $11.3 billion a year earlier.

HEADS UP

ON THE HORIZON

MEERKAT ON YOUTUBE

The Claude Situation Just Got WORSE... (Anthropic IPO). Anthropic's sales document says its own product could pose existential risks, and spends 80 of its 261 pages on what could go wrong. Meerkat explains why the scariest prospectus yet changed almost nothing. Watch on Meerkat Explains.

THE BURROW

Quick vote. The FTC is investigating OpenAI and Anthropic over AI agents that went rogue. What do you make of it?

Trivia. The Federal Trade Commission, now investigating OpenAI and Anthropic, was created in 1914 under which US president? Answer's at the bottom of the email. No peeking.

ALSO WORTH YOUR INBOX

The daily crypto brief from the same burrow. Crypto Nutshell is Bitcoin, the news and the numbers in a five-minute read.

HOW DID WE DO?

One tap, and Meerkat listens. How did today's issue land? Hit a button, and add a line if you want to tell Meerkat why.

Jobs report tomorrow at 8:30. Meerkat is keeping the humans' numbers and the machines' paperwork in separate piles.

Got a friend who reads every line of their tax return? Forward them this.

Meerkat is not a financial adviser, and nothing here is advice. Meerkat is a small animal with a whiteboard and a subpoena it hasn't opened yet. Please do not sue Meerkat.

TRIVIA ANSWER

Woodrow Wilson. He signed the Federal Trade Commission Act in September 1914, the same year as the Clayton Antitrust Act.