Quartz, a well-known business news outlet, has quietly started publishing articles written by artificial intelligence under the name “Quartz Intelligence Newsroom.” These pieces, which summarize reporting from other outlets like CNN and The Associated Press, are appearing more frequently, with 18 already live as of this week.

Each AI-generated article is around 400 words long and lacks direct quotes from sources. Instead of attributing information within the text, Quartz’s AI simply lists its sources at the top of the article. While this approach keeps things brief, it raises questions about how much detail and context might be lost in the process.

A spokesperson for Quartz’s parent company, G/O Media, confirmed that this is an “experimental” project. The goal, they say, is to free up Quartz’s human journalists to focus on more in-depth reporting. However, the quality of these AI-generated articles has already come under scrutiny. For example, one piece summarizing a TechCrunch article on deleting Facebook, Instagram, and Threads accounts offered vague instructions that lacked the step-by-step guidance of the original.

Quartz isn’t the first media company to experiment with AI-generated content. CNET, Gannett, and even Sports Illustrated have tried it, often with mixed results. For publishers, the appeal of AI lies in its cost-effectiveness—no salaries or benefits required. But the trade-off is often accuracy and depth, as seen in some of Quartz’s clunky headlines and contradictory statements.

Despite these challenges, G/O Media claims reader engagement with the AI stories has “far exceeded expectations.” The company also denies any financial struggles, stating it’s “very well funded” and even hiring more editorial staff. Still, as AI continues to creep into newsrooms, the debate over its role—and its impact on journalism—is far from over.