America's biggest newspaper chain published AI sports write-ups with the team names left as blank tags, and paused the tool

By Ellie Harris · Filed Botched recaps published and paused August 2023

Alleged: Gannett; AI tool from LedeAI developed or deployed the AI system implicated in this incident. Details are drawn from public reports; parties are presumed innocent of any wrongdoing not established by an official finding.

What happened

It was reported that in August 2023, Gannett, the largest newspaper publisher in the United States, began using an AI tool from LedeAI to write short reports on high school sports.

The experiment quickly ran into problems.

Reporting indicated that some of the published stories still contained placeholder text instead of team names, leaving bracketed tags where the names should have appeared. Others repeated the same sentences, used awkward language, or restated basic facts multiple times, making it obvious the articles had not been properly checked before publication.

The errors were widely reported online and quickly became an example of AI-generated content being published without adequate human review.

Gannett said it was experimenting with AI to support its journalists and was evaluating different tools. After the errors attracted public attention, it paused the use of LedeAI’s automated sports stories. LedeAI acknowledged the problems, including the placeholder text, repetition and awkward phrasing, and said it was working to improve the system.

What an auditable version would have shown

An auditable publishing system would show whether each story had been reviewed before publication. It would record what the AI generated, what an editor changed, who approved the final version, and whether automatic quality checks detected problems such as placeholder tags, repeated sentences or incomplete text. Instead of readers discovering the mistakes, the system would have caught them before the stories went live.

Where the gap was

The problem wasn’t that AI wrote the articles. The problem was that obviously unfinished copy reached publication. A tool can generate hundreds of stories in minutes, but speed is no substitute for editorial review. There was nothing in the publishing process preventing clearly broken output from being treated as finished journalism.

What governance should have looked like

Any AI-generated article should pass through a human editor before it carries a newsroom’s name. Basic quality checks should automatically stop stories containing placeholder text, repeated paragraphs or incomplete fields before they can be published. A VerificationGate is designed to prevent unfinished AI-generated copy from reaching readers until a person has reviewed and approved it. A ConductRecord keeps a record of what the AI produced, what was edited, and who approved publication. The goal isn’t to stop news organisations using AI. It’s to make sure readers don’t become the final quality check.

The reference implementation of VerificationGate and ConductRecord is open source. It lives at github.com/saffronandindia/headlights-oss, Apache 2.0 licensed, free for any company to install. The repository is public now.

Sources

The mailing list

Fresh incident reports every week. One email to match.

We add new incidents to the library regularly, and send a single short email each week with what's new. The library stays free and open; this is just how you keep up with it.

No tracking. Unsubscribe in one click.

The record

An auditable system would have produced a signed, tamper-evident record the moment this happened: what the system did, the version that did it, the basis it acted on, and the action taken, and Gannett; AI tool from LedeAI could have produced it on demand.

This is the record the system as deployed did not produce in a signed, auditable form.

What this teaches

Capture what happened when it happens

What the system did, the version that did it, the basis it acted on, and the action taken, recorded at the moment, not reconstructed after.

Sign it, so no one has to trust the record-keeper

A tamper-evident entry. Edit it later and the signature breaks. The record does not ask for the benefit of the doubt.

Make it verifiable by anyone

A court, a regulator, a customer's lawyer can check the record themselves, without taking the company, or us, at our word.

Also in the library

HD-INC-001 Air Canada chatbot promised a bereavement refund policy that did not exist Aviation · 2022 HD-INC-002 Mata v. Avianca, the lawyer who cited six cases that did not exist and asked ChatGPT to confirm them Legal services · 2023 HD-INC-003 Michael Cohen gave his lawyer fake case citations he had got from Google Bard, and his lawyer filed them in a federal court Legal services · 2023

Headlights summarises publicly reported AI incidents. All summaries are independently written, attributed to their original sources, and intended for research and educational purposes. Allegations are identified as such until established through official findings.

This report is based on the sources listed above and reflects information available at the time of review; later developments may not be captured. Where a person is described as charged with or alleged to have done something, that allegation is unproven unless a conviction or a court or regulatory finding is stated. Headlights publishes journalism and commentary, not legal advice.

Want to write back?

Direct to my inbox.

ellie@useheadlights.com →