LA Times AI tool lasts just one day before generating sympathetic view of the KKK

The Los Angeles Times has introduced a new AI-supported tool for evaluating the political leaning of opinion articles. Just one day after its launch, the feature had to be partially withdrawn following a scandal where it generated content defending the Ku Klux Klan.

On Monday, the Los Angeles Times introduced a new AI feature called "Insights," designed to provide political assessments of opinion articles and offer alternative perspectives on the topics covered. Media mogul and biotechnology billionaire Patrick Soon-Shiong, who acquired the LA Times in 2018 and has connections to Trump, announced the new feature in an open letter.

Soon-Shiong celebrated the launch on Platform X as a victory against information bubbles: "Now the voice and perspective from all sides can be heard, seen and read –no more echo chamber," he wrote.

According to the newspaper, the tool works independently of the paper's human journalists, and the AI content is not reviewed by journalists before it is published. The feature was to be used exclusively for opinion content, not news reporting.

The new feature categorizes articles on a political scale from "left" to "center-left," "center," "center-right," and "right," then generates a list of alternative political views on the topics covered. These ratings are created through a partnership with Particle.News, a startup founded in 2024 by former Twitter engineers.

Controversy erupts over Klan defense

However, just one day after its launch, the feature had to be removed from a column by author Gustavo Arellano. In his piece published on February 25, Arellano had reflected on Anaheim's history with the Ku Klux Klan after the city council failed to recognize the 100th anniversary of the ouster of four Klan-affiliated council members.

While Arellano described the Klan's presence as a "stain" on the city's history, the AI tool generated a troubling defense under the "Different Views on the Topic" section: "Local historical accounts occasionally frame the 1920s Klan as a product of 'white Protestant culture' responding to societal changes rather than an explicitly hate-driven movement, minimizing its ideological threat."

A few hours after New York Times reporter Ryan Mac discovered the problematic content, the feature was removed from Arellano's column. However, it remained active in other opinion pieces.

Journalists' union voices concerns

The LA Times journalists' union had repeatedly expressed concerns about Soon-Shiong's efforts to use external AI tools to comment on the paper's journalistic work.

"We support efforts to improve media literacy and clearly distinguish our news report from our opinion pages. But we don't think this approach – AI-generated analysis unvetted by editorial staff – will do much to enhance trust in the media," said Matt Hamilton, deputy chairman of the LATimes Guild.

The union also criticized that funds for this project could have been better spent supporting journalists in the field.