Das Magazin
21 September 2026
After years of pilot projects, newsroom AI tools are moving from experimentation to everyday use. New surveys show a majority of journalists are using AI for transcription, translation, and copy-editing, while about a third have integrated it into research and story structure. But despite AI's gradual expansion, most publishers have not yet used it to generate full first drafts of articles.
The Reuters Institute's Journalism and Technology Trends and Predictions survey of 280 leaders from 51 countries found that about 80% expect their newsrooms to use AI more in the future. The Associated Press survey of 300 journalists found that 70% of organizations have used generative AI, with common applications being social media posts, newsletters, headlines, and transcribing interviews.
But when it comes to using AI to actually write a full first draft, newsrooms are more cautious. While a majority of journalists are gradually incorporating AI for supporting workflows, only a fraction are using it for core editorial tasks — and even fewer are letting it generate full articles on its own.
The latest data shows a widening split between the growing adoption of AI as a support tool, and its more tentative use in writing editorial work itself. Most journalists now treat AI as an integrated part of their daily toolkit, but editorial leaders remain watchful about applying it to the core work of shaping and fact checking a news story.
In the UK, a Reuters Institute survey of 1,004 journalists found that 56% are using AI at least once a week, but only 15% see it as a major opportunity. Two thirds say it's a significant threat. That caution also appears internationally. While the Reuters Institute survey shows that about 80% expect increased AI use, most still view AI as more of an efficient tool rather than an editorial partner.
From what newsrooms are or are not doing with AI, you can see the split. The AP survey found that 70% of organizations have used generative AI, but most common uses were for social media, headlines, and transcribing interviews, not drafting full stories. Similarly, a WAN-IFRA survey reported that half of newsrooms are using AI, but only 32% had used it for article creation. And while a Muck Rack survey found 28% of journalists using or planning to use AI, it lists uses like brainstorming and research, but not full draft generation.
For the kinds of tools journalists use most often, AI has made inroads, particularly for what are essentially its facilitative uses: transcription, translation, and copy-editing. The Reuters Institute survey found 49% used it at least monthly for transcription, 33% for translation, and 30% for grammar checking and editing.
But when it comes to using AI to help with research, idea generation, or writing, a smaller fraction are using it at least monthly — 22% for research, 16% for brainstorming and generating headlines, and 10% for generating first drafts. That shows that while many journalists are experimenting with using AI for core editorial processes, most have not made it part of their regular workflow.
The numbers are just one part of what's shaping up to be a sluggish, organically expanding use of more advanced AI in newsrooms. Leaders are watching closely to see when and how it can support their staff's work, while watching for any red flags on the output. It seems that with generative AI, most publishers are aiming for the same kind of support-system combination: broad adoption as a workflow tool, along with cautious eyes on its editorial role.
