Artificial Intelligence is no longer a side project in newsrooms — it is the new operating system across the media business. That was the message as INMA gathered 180+ delegates from 26 countries at San Francisco’s KQED studios for its Media Tech & AI Conference last week.

Five themes stood out during the two-day conference:

• AI as the new operating system of news media: The news industry’s next platform shift, transforming every layer from content creation to monetisation.

• Economics and power in the AI era: Rewriting value chains and forcing publishers to reclaim control over data, rights, and revenue.

• Content rights and licensing frameworks: Building machine-readable, monetisable systems like RSL to ensure fair compensation in the AI ecosystem.

• Data governance and ownership: Success in AI hinges on trusted, well-governed data and full control of tech stacks.

• Redefining discovery and distribution: Answer engines and conversational AI are reshaping how audiences find and engage with news.

While journalism threaded throughout the conference programme, an unexpected visit by California Governor Gavin Newsom — considered the top candidate for the Democratic Party for the next U.S. presidential election — reminded delegates of the news industry’s North Star. He encouraged news media to keep “shining a bright light on abnormal political conditions,” adapting to new technology and influencers to bolster the industry’s civic mission. 

Newsom’s INMA conference appearance underscored a Media Tech & AI Week — conference and study tour — that blended urgency with optimism about AI’s role in strengthening democracy and sustainability.

Power and economics in the AI era

James Cham, partner at Bloomberg Beta, set the conference tone by reframing the AI debate as an economic one. 

“The one-to-many model is dead,” he said. “For decades we created one product for millions of people. When creation becomes free and distribution becomes smarter, everything changes and CPA becomes more important than CPM.”

Cham warned news publishers not to assume that partnerships with major tech firms will automatically preserve their value: “There’s this sense that media companies will feed all this data to someone else and maybe get a few pennies. That’s probably wrong-headed.”

He pointed to the mid-1990s New Century Network — an early alliance of U.S. newspapers — as a lesson in misplaced confidence. 

“They focused on commercial terms before they understood the new medium,”Cham said. “We can’t repeat that mistake.”

His challenge: Stop relying on augmentation alone. “The parts that transform your organisations are the parts that are 100% automatable,” he said, urging news publishers to build internal experiments rather than wait for perfect vendor solutions.

Cham’s core reassurance was that no one — not even Silicon Valley, where some attendees participated in a study tour earlier in the week — has all the answers: “Everyone is figuring it out right now. The big companies are powerful, but they’re not as powerful as you think.”

Redefining innovation and imagination

Karen Silverman, CEO of the Cantellus Group, widened the lens beyond media: “If you don’t change direction, you may end up where you’re going,” she said, quoting a favourite line that drew murmurs of recognition across the room.

Her argument: Leaders must balance fundamentals and imagination. Fundamentals include data quality, governance, and processes robust enough to handle AI-scale workflows. Imagination means the space and courage to rethink what a news organisation is for.

“Everybody wants to do something,” Silverman said. “Everybody also knows they can’t do everything. There’s a real premium on figuring out what’s the smart thing to do.”

Drawing lessons from transportation, finance, education, and health care, she highlighted that the most successful sectors paired human insight with machine capacity rather than replacing one with the other. 

“The organisations pushing new technology through old pipes are going to break them,” she warned.

Discovery, distribution, and the changing value chain

A major theme across the conference sessions was the redistribution of traffic and trust as AI alters discovery. Chartbeat CEO John Saroff presented data showing how answer engines and on-site chat tools are reshaping engagement: “What you measure is going to change because how audiences arrive and interact is changing underneath you.”

Google’s Danny Sullivan outlined ongoing efforts to balance AI-driven search with open-Web visibility. Meanwhile, Andi Search CEO Angela Hoover described a new generation of conversational engines. “People don’t want paragraphs cited to them,” she said. “They want an expert friend who knows them.”

News companies, she argued, should create content structured for context — not just for clicks.

The conversation deepened in a panel featuring AP, Perplexity, and OpenAI. Sara Trohanis of AP explained how structured data licensing could model fair compensation; Jessica Chan of Perplexity stressed transparent attribution; and Christina Lim of OpenAI said cooperation with publishers is central to building trustworthy systems.

Microsoft’s Nikhil Kolar revealed progress on an AI content marketplace to pay publishers for training data. Doug Leeds and Eckart Walther introduced Really Simple Licensing (RSL) — a machine-readable protocol allowing news publishers to declare how their material can be used. Together, these initiatives suggested the long-promised “content payment” framework for the AI age may finally be forming.

Building competitive advantage through data and design

From content economics, the discussion moved to the foundations that enable AI success. 

Ingrid Verschuren, executive vice president of data intelligence and AI governance at Dow Jones, said, “AI is only as strong as the data that fuels it.”

Her road map emphasised governance and culture over technology: “It’s not just about storage and scale — it’s about human expertise at every stage.” She described Dow Jones’s evolution from data collection to “data intelligence,” treating information as both asset and responsibility.

Vineet Khosla, chief technology officer at The Washington Post, linked that idea to brand strategy. “Journalism has to remain essential in an algorithmic world,” he said. “To do that, we must own our technology stack and our data.”

In product innovation, Beyond Words founders Patrick O’Flaherty and James MacLeod showed how AI is transforming audio into a discovery layer for publishers. “Voice is becoming a user interface,” O’Flaherty said.

Complementing that vision, Adam Mosam of Channel 1 demonstrated multi-modal storytelling that merges text, video, and voice — a reminder the next platform war will be fought across formats, not devices.

Agentic AI and automation on the horizon

If the first wave of AI was about assistance, the next may be about autonomy. Raphael Daste of Stripe, Cosmin Ene of Supertab, and Toshit Panigrahi of Tollbit explained “agentic AI” — systems that act independently to personalise, distribute, and even transact.

Daste warned governance must evolve just as quickly as capability: “As soon as an agent can act, it can also err,” he said. Yet the commercial opportunity is vast: “Agentic commerce could turn every content touchpoint into a revenue event.”

At a working lunch during the conference, the Lenfest AI Collaborative — joined by The Philadelphia Inquirer, Boston Globe Media, The Seattle Times, and The Dallas Morning News — showed how collaboration can accelerate experimentation. Chris Patheiger from Dallas summed it up: “Duplication is the enemy. Collaboration lets us move ten times faster.”

Talent, risk, and resilience

Across later sessions, executives focused on what AI means for organisations themselves. Katharina Neubert of Holtzbrinck said, “AI isn’t just changing how we work — it’s changing who we need.” She advocated for hybrid skill sets that combine editorial sensibility with product fluency.

Will Allen of Cloudflare demonstrated how the same tools driving efficiency can also amplify threats. His vivid examples of deepfake attacks and phishing campaigns led to a sober warning: “Treat cybersecurity as a newsroom function, not just an IT one.”

Benedict Nicholson of Newswhip showed that while social referrals have softened, meaningful engagement remains high when brands build trust and conversation rather than volume.

Finally, Ben Gerst of Nota, joined by Jeff Moriarty of Nexstar Digital, showed how AI infrastructure is improving productivity without eroding standards. “Assistive AI isn’t about cutting corners,” Gerst said. “It’s about protecting quality while scaling output.”

Leading with purpose

Throughout the conference, moderators INMA Product and Tech Initiative Lead Jodie Hopperton (who curated the entire week), INMA Digital Platforms Initiative Lead Robert Whitehead, and Morgan’s Chief Technology Officer Nicki Purcell anchored discussions with a single refrain: Act deliberately, not reactively. 

In the closing podcast recording, Hopperton said, “AI isn’t a side project anymore. It’s the next platform shift — and the publishers who experiment, share, and set standards together will lead it.”

Silverman had offered a similar provocation earlier in the sessions: “Leaders who aren’t asking questions about the roles of humans and technology will have those questions answered for them — by someone else or by something else.”

That was perhaps the deepest message of the week. The future of news will not be built by waiting for platforms or regulators. It will be built by news publishers who understand that intelligence — human or artificial — is now the core raw material of their business.

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