Analyzing the News: The AI spending sell-off

Twelve outlets covered the tech sell-off over the cost of AI. Fíor read two of them as AI.

Fíor is a natural language classifier that predicts whether a piece of text was more likely written by a person or by AI.

On 22 October, the Nasdaq fell 2.6%, its steepest drop since early June. Meta was at the centre of it. Its results beat expectations, but it reported its first cash outflow in more than a decade, narrowed its 2026 spending plans to between $135 billion and $145 billion, and announced a new class of non-voting shares it could use to raise money. Its shares fell about 7%. Chipmakers followed, and Nvidia's chief executive, Jensen Huang, warned that capital to fund AI infrastructure was getting scarce. Rising oil prices added to the pressure.

The question running through the day was no longer how fast AI can grow, but who pays for it and when it pays back. It is the second time in a fortnight that AI has driven a sell-off. We covered the first one with Fíor's previous model, so the two sets of results are not compared here.

The outlets were Reuters, AP, the Wall Street Journal, Business Insider, Al Jazeera, Yahoo Finance, Kiplinger, CBS News, Forbes, Fortune, the New York Post and The Telegraph. That is two wire services, American and British newspapers, the business and personal finance press, and Al Jazeera. The Wall Street Journal sits behind a paywall, so only part of it was scanned. The Telegraph was scanned as syndicated on AOL. CBS News says AP contributed to its report. Yahoo Finance's coverage was a live blog.

The result

Ten read as human: Reuters, AP, the Wall Street Journal, Business Insider, Al Jazeera, Kiplinger, Forbes, Fortune, the New York Post and The Telegraph.

Two read as AI: Yahoo Finance and CBS News.

The Wall Street Journal read as human, but not confidently, on the shorter sample its paywall allowed. CBS News read as AI, also not confidently, and Brath said it reads like a person wrote it. Yahoo Finance was the clearest AI result in the set, and both of Brath's labels agreed.

This is the seventh story scanned with Brath, Fíor's new model. CBS News has now read as AI in three stories running. AP, which contributed to the CBS report, read as human.

The one that stood out

Yahoo Finance's live blog read most clearly as AI. Live blogs are the obvious explanation: fast, updated through the day, stitched together from many small pieces.

They do not explain it on their own. In the Google Cloud outage, three live blogs read clearly as human.

The most templated article in this set points the same way. AP's piece is a short daily item listing where each index closed, written to the same pattern every trading day. It read as human.

Why market coverage reads like this

The daily market report is one of the most formulaic forms in journalism. Index levels, points, percentages and the same few company names, published every trading day by almost every business outlet. If any coverage were going to read as AI, it would be this.

Most of it did not. The pieces that went beyond the numbers read as human too. The Telegraph built its story around Huang's warning on capital. Al Jazeera asked whether the spending amounts to a bubble. Forbes set Meta's spending against Tesla's cut.

Some of these pieces may have had AI help, with a first draft, a rewrite or a final polish. Or market coverage has become formulaic enough that a reporter and a model would write it the same way. Fíor cannot tell those apart on its own.

Investors spent the day asking whether the money going into AI will ever pay back. The reports on that question were mostly the most routine writing in journalism, and almost all of them read as human. Fíor can see the shape of a piece of text. It cannot see how a newsroom put it together.

What Fíor is

Fíor is a machine learning classifier, currently in beta and available as a browser extension. It gives a prediction, not a verdict. It cannot see who sat at the keyboard, and nothing here is a claim about any journalist or publication. Reading as AI and being written by AI are not the same thing.

Treat it as a signal, not proof, and check the pages you read yourself.

Try the browser extension at Fíor.

Analyzing the News is a series where we run one major story through Fíor and see what comes back.