Skip to content
Video Microsoft on YouTube Mar 2026

Microsoft: How USA Today uses AI in the newsroom

This video, produced by Microsoft and published in March 2026, features a USA Today editor walking through how the publication has integrated AI into its newsroom operations. The format is a practitioner walkthrough — an actual editor describing decisions made in actual coverage — rather than a vendor demonstration. The Microsoft framing is present but does not dominate the content.

USA Today’s scale makes the use case interesting. The publication produces a large volume of content daily across dozens of topic areas, operates on a tight publishing cadence for breaking news, and serves a wide demographic range. The AI integration is not about replacing editorial work in a boutique publication — it’s about managing volume and speed without losing editorial control in a very large one.

The editor covers three areas of the workflow that AI has changed.

The first is breaking news coverage. When a story breaks, USA Today needs to publish quickly, which historically meant publishing early versions that were thin on context. The workflow now uses AI to pull contextual background — prior coverage, relevant data, related events — into the drafting environment as a story develops. Editors can access that background without leaving the drafting tool, and the system updates it as new information comes in. The goal is not to automate breaking news; it’s to make sure the first published version has adequate context.

The second is content personalization at the homepage and newsletter level. USA Today sends a large number of newsletters, and manually curating each version at scale was not sustainable. The current system uses engagement data to adjust story selection and ordering for different audience segments. Editorial judgment still governs which stories are in the pool; the AI handles prioritization for different audiences.

The third is daily production tasks: headline variants for A/B testing, article tagging for SEO and content recommendations, and caption generation for photo assets. These are tasks that don’t require editorial judgment but consume significant time when done manually. Automating them freed up time that editors describe as going into work that does require judgment — fact-checking, source development, deeper reporting.

The editor is candid about what has not worked: fully automated summaries of complex investigative pieces, AI-suggested story angles that don’t account for the publication’s existing coverage of a topic, and any task where the AI output requires as much editing as starting from scratch.

Worth watching if you are responsible for content operations at a publication with high volume, multiple channels, or a daily production schedule that makes manual handling of routine tasks unsustainable. The USA Today use case is at a scale most publications don’t match, but the categories of tasks they have automated are relevant at smaller scale too.