AI is no longer a futuristic concept, but a present-day imperative, according to Sonali Verma, INMA Generative AI Initiative lead.

“Everyone is using AI. It’s no longer a nice-to-have,” Verma said at this week's said at the Generative AI Town Hall. “It’s very much a must-have if you want to compete against everyone else in the news business.”

During the town hall, media leaders from HT Media, Der Spiegel, and USA Today Co. shared how their companies are leveraging AI in the newsroom and beyond while maintaining editorial control and the trust of their audiences.

HT Media solves everyday problems

There are multiple reasons why AI matters in the newsroom, Abhishek Sharma, HT Media's chief product officer, said: “There is no greater time than now a need for multiple formats to be offered.”

The company has been building on four AI-powered products: RankAI, DeskAI, Content Formats, and Echo. The first two products, RankAI and DeskAI, are used primarily on the editorial side.

RankAI, an SEO automation product, consists of four core AI agents: Trends Agent, Search Agent, SEO Comparator Agent, and UI Analysis Agent.

The Trends Agent compares parameters — such word counts and image quality — to competitors’ parameters and then gives actionable recommendations to the newsroom.

The impact has been notable. SEO coverage has improved by 50%, and user traffic has improved by 20-25%.

“We’ve been able to identify more accurate ranking insights, which can help us in improving our article generation to rank on those trends or keywords,” Sharma said.

DeskAI is a newsroom AI editorial engine which aims to offer solutions to four everyday problems:

• AI draft creator: Auto creates first draft on any topic, recent or archive
• Graphic creator: Converts any story into visual cards, charts, social-ready graphics
• PDF to story converter: Turns earnings reports, documents, structured data into drafts
• This day, that year: Identifies key events on today’s dates and suggests topics for newsroom

“What has happened with these four products, which have been used by different news desks inside the newsroom, is that our time to publish quality stories for deep explainers, deep research stories, has reduced by 40%,” Sharma said, adding that in addition to reduced publish time, the content being produced is filled with twice as many multi-media formats such as graphic and data charts.

Der Spiegel personalises without commodifying

In Der Spiegel’s bustling newsroom, the question isn’t whether to use AI — it’s how to do so responsibly.

Ole Reissmann, director of AI at Spiegel-Gruppe, said the team has tackled a question plaguing publishers everywhere: how to personalise without commodifying.

Rather than rushing to full automation, Der Spiegel is building an AI architecture step by step. The first feature in development: automated article summaries.

“We are a bit late to the party, but we really want to get this right,” Reissmann said. Behind that caution lies an ethical tension. “AI can write a summary — but then, do you want to let editors approve each and every summary? And if an editor approves an AI summary, do we have to label it as AI content?”

He admitted to feeling “very conflicted about this,” concluding that when human oversight and judgment are present, the output remains fundamentally human work: “Then it’s just a tool,” he said, “and we don’t have to label it as AI — or do we?”

With products like ChatGPT’s Plus feature, which “crunches all the things you have asked the other day” and delivers a morning news briefing, AI platforms are already becoming news aggregators.

“Discovery, personalisation — all that stuff is happening on AI platforms,” he said. The question for publishers: “Can we build a great user experience? … Or do we have to do something differently?”

In the face of those forces, Der Spiegel has made a deliberate choice. “I know that for some, it might be a way to produce more content faster, more cheaply,” Reissmann said. “We didn’t choose that route.” Instead, the strategy is to “double down on our brand” and “put humans front and center — our readers, for sure, but also our editors and writers.”

Gannett expands and deepens its reach

When it comes to actual newsroom tools, Jessica Davis, vice president of news automation and AI product at USA Today Co. (formerly Gannett), highlighted two practical, high-impact examples: AI-assisted article summaries and community coverage automation.

The first, called Key Points, uses a fine-tuned Gemini model integrated into Gannett’s content management system. “Journalists can choose to get an AI system summary after they write their article,” Davis explained. “They review it… they can change it, they can ditch it, whatever they want to do.”

These AI-generated summaries have proven not only safe but successful.

“We found as long as the key points were more of a teaser, they were very effective in actually improving engage time,” Davis said. “We saw some pretty incredible jumps in engage time for some of our pilot newsrooms.”

She also dispelled one major myth: “There was a lot of fear about what Google would think about AI-assisted summaries. And we have not seen that it has hurt our search performance.”

The second case study — AI-assisted reporting — is even more local.

“Especially in our smallest newsrooms, we are able to bring hyper-local, micro-level news to communities in greater quantities, in a more cost-effective way than we’ve ever been able to do before,” Davis said.

That means using AI to help format community announcements or press releases, always verified and edited by journalists.

“We call them AI-assisted reporters for now,” she said. “The journalist checks it, another editor checks it, and they publish it. We haven’t had any errors to date, and the response from the community has been so positive.”

Davis emphasised that these projects aren’t just about efficiency — they’re about expanding journalism’s reach and resilience. “This is a real positive story of bringing local news in,” she said. “Other teams are exploring how this could help in different-sized markets — things like events calendars and community submissions.”

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .