Analyzing the News: DoorDash buys Instacart

Twelve outlets covered DoorDash's $17 billion deal for Instacart. Fíor read three of them as AI, and one of them for the sixth story running.

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

On 29 October, DoorDash agreed to buy Instacart for $65 a share, about $17 billion. That is a 26% premium to Instacart's share price before reports of the talks emerged, and at least three quarters of it will be paid in cash. The deal would bring together the biggest name in American restaurant delivery and the biggest in grocery delivery. Both companies plan to keep their brands and apps, and to combine their membership schemes. The deal is not expected to close until the end of 2027 or the first half of 2028, and it needs regulators' approval first.

The companies pitched it as a way to compete with Amazon, Walmart and Uber. Critics, including the anti-monopoly group American Economic Liberties Project, warned it could mean higher prices and less choice for shoppers, delivery workers and local shops.

The outlets were Reuters, AP, Bloomberg, the BBC, TechCrunch, the Los Angeles Times, CBS News, NBC News, ABC News, CNN, Forbes and the New York Post. That is two wire services, American and British broadcasters and newspapers, the business press and TechCrunch. NBC News was scanned as published by NBC Chicago, CNN through a partner station and Forbes through Yahoo Finance, each under its original byline. The Los Angeles Times sits behind a paywall, so only part of it was scanned.

The Washington Post was left out. In the last three stories, its scan picked up an AI-written summary of reader comments, and this time it was not confirmed that the scan avoided it.

The result

Nine read as human: Reuters, AP, Bloomberg, TechCrunch, NBC News, ABC News, CNN, Forbes and the New York Post.

Three read as AI: the BBC, the Los Angeles Times and CBS News.

Two of the three were splits. Brath judged the BBC and Los Angeles Times articles likely written by AI, but said both read like a person wrote them. For CBS News, it was unsure how the piece reads. None of the pages drew a writing-habit flag.

This is the fourteenth story scanned with Brath, Fíor's new model.

CBS News, six in a row

CBS News has now read as AI in six stories running where it ran its own reporting: the PHP security flaws, Meta's New Mexico penalty, two AI sell-offs, Anthropic's IPO filing and now this. The two times it read as human in between, it was carrying copy from AP or AFP.

That is the most consistent result in the series. It is worth being careful about what it means. It says that one model, reading the opening of CBS News articles, keeps finding the same patterns. It does not say how CBS writes them. A house style, a template for business news, editing tools or AI help could all produce the same result, and Fíor cannot tell those apart.

Why deal coverage reads like this

Merger stories start from a company announcement, with the price, the premium and the closing date already written out. Most outlets reported the same terms in much the same order.

Where they differed was in who they thought the deal was for. Reuters and Bloomberg wrote about competition with Amazon and Walmart. The BBC and ABC News led on what shoppers would get. CNN, NBC News and the Los Angeles Times gave more room to workers and local businesses. The New York Post led on the risk of higher prices. The angle did not predict the result. The BBC and ABC News took similar angles, and one read as AI and the other as human.

Some of these pieces may have had AI help, with a first draft, a rewrite or a final polish. Or deal 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.

How much of grocery delivery would the combined company control? CNN cited one estimate of about 60%. The New York Post cited another of 88%. Both can be right, because they count different markets. Fíor's results need the same care. Six in a row is a real pattern, in what one model measures, and nothing more than that.

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.