From Palo Alto to downtown San Francisco, news executives met the technologists, founders, and researchers who are quietly rebuilding the foundations of the information economy.

What they found was both exhilarating and unsettling — a future where the competitive edge for news publishers will not only depend on unique, highly verifiable content creation but also on how they design trustworthy systems, govern data, and weave AI into every layer of their organisations.

As part of its Media Tech & AI Week, which drew more than 180 delegates from 26 countries, INMA took 40 news professionals to 14 tech and AI companies in Silicon Valley and the San Francisco Bay Area, peeling back the curtain on a new media reality: Artificial Intelligence is no longer an add-on; it is the augmenting system of modern journalism. 

Study tour participants traced the contours of this transformation across companies as different as Otter.ai, Cloudflare, Google, and OpenAI. Each stop revealed how fast AI has shifted from innovation to day-to-day: 

• Start-ups that once transcribed conversations now create knowledge engines.

• Cybersecurity firms stake claims to become stewards of digital ethics.

• Tech giants are remapping some of the ways through which audiences discover news. 

The cumulative message: AI is redefining how journalism is produced, distributed, protected, and valued.

If the technology itself dazzled, the human implications ran deeper. Executives spoke less about machine capabilities and more about leadership: 

• How to preserve editorial integrity in automated systems.

• How to make data transparent and accountable.

• How to guide teams through cultural change. 

Across conversations, study tour participants heard that provenance, explainability, and collaboration will become the hallmarks of credible media in an AI-driven world.

By the time the group gathered at San Francisco’s waterfront to close the tour, the mood was reflective. The tools, they agreed, are already here; what matters now is stewardship. The next chapter of AI in news media will be written not by coders alone but by leaders capable of balancing innovation with conscience.

AI becomes infrastructure

Across the three-day study tour of Silicon Valley and the San Francisco Bay Area, one message was clear: AI is no longer an experiment. 

From transcription platforms to cybersecurity labs, tech executives told INMA study tour participants that AI now powers the invisible systems underpinning communication, knowledge, and decision-making — and that publishers who treat it as a bolt-on feature risk being left behind.

At Otter.ai, founder Sam Liang described how his company’s mission evolved from transcription to knowledge creation. “Our mission is to make conversations more valuable,” he said. “People have been talking for thousands of years, but most of that voice data is lost.” Otter built its own speech-recognition engine to turn dialogue into searchable, shareable intelligence — the same logic newsrooms can apply to interviews, editorial meetings, and source material.

Liang’s next step — “AI meeting agents” that can participate in discussions — illustrated a deeper shift: automation moving from recording to reasoning. For publishers, this signalled that newsroom memory, not content generation, may be the most transformative use of AI.

The new value of data

If Otter showed the power of capturing internal knowledge, Stanford University’s Cheryl Phillips demonstrated the value of sharing it. Through her Big Local News project, she has built open datasets to support accountability reporting worldwide. 

“When you work on a big investigation, you collect a wealth of data that too often gets left on the cutting-room floor,” she said. “I wanted to make sure that data could be used again — by other journalists, in other places.”

Her work — from tracking police stops to mapping school-enrolment declines — models a future where data becomes common infrastructure. Participants saw parallels to first-party data strategies in publishing: collecting responsibly, pooling for collective benefit, and layering AI tools that make data usable through natural-language queries.

At the same time, companies like Cloudflare illustrated the other side of the data coin: protection. Study tour participants were told that as models grow, scraping of publisher sites happens at scale and often without consent. Cloudflare’s answer is transparency and control — real-time blocking of unauthorised bots and a “pay-per-crawl” pilot that could create a market for licensed AI training data.

For publishers, the shift was striking. Where once audience reach defined digital success, the new currency is data integrity: knowing what information you hold, who uses it, and how value returns to your organisation.

Generative ecosystems and distribution

A visit with OpenAI underscored how quickly those economics are changing. Opportunities for media organisations are evolving from one-way syndication to two-way collaboration inside conversational ecosystems. Delegates heard how ChatGPT apps could open a direct channel between audiences and publishers — allowing media brands to deliver personalised, authenticated content inside AI environments rather than through traditional referral traffic.

The same redefinition of discovery was evident at Microsoft’s offices. Aparna Lakshmi Ratan, partner product director for Microsoft AI, explained how MSN’s algorithms now use large-language-model reasoning to “improve discovery, relevance, and engagement” — a model that rewards publishers producing authoritative journalism optimised for clarity and trust.

Meanwhile, LinkedIn’s Catherine Taibi positioned the professional network as an emerging top-of-funnel for credible news: a place where executives seek perspective rather than headlines. Its evolving mix of verified voices and newsroom partnerships is helping publishers reach high-value audiences as search traffic declines.

At Google’s Mountain View campus, study tour participants heard how SythID and C2PA can help with responsibly disseminating information. Participants also heard how Google Trends can be used as an indicator and how queries are getting longer – demonstrating changing consumer habits. Gems and Notebook LM are two Google tools for news organisations. 

Rounding out the picture, Vermillio co-founder Dan Neely introduced TraceID, technology that can detect how text, images, and even writing styles are used within AI models. For publishers, it’s a blueprint for content-licensing accountability.

Together these companies outlined a single conclusion: The future of distribution is traceable. Publishers that know where their journalism travels — and how it powers other systems — will hold leverage in negotiations that follow.

Transparency and responsibility

At cybersecurity start-up CTGT, founder Cyril Gorlla described how his company builds enterprise platforms to make AI models faster, safer, and more interpretable. Its technology lets companies customise and deploy foundation models with their own data and rules, monitor outputs in real time for bias or hallucinations, and remediate issues automatically. Targeting regulated sectors like finance and health care, CTGT emphasises efficiency — training models up to 10 times faster with far less compute — and transparency from first principles rather than black-box fixes. As Gorlla puts it, trust in AI depends on making its inner workings inspectable and controllable.

That insistence on auditability echoed through later sessions with Hearst Newspapers including the San Francisco Chronicle. Senior executives Emilio García-Ruiz, Tim O’Rourke, and Derrick Ho explained how they are testing AI tools across newsrooms of vastly different sizes while keeping human oversight central. The challenge, they said, is cultural as much as technical: encouraging experimentation without compromising ethics or accuracy.

San Francisco Chronicle Publisher and CEO Bill Nagel added that leading a company through AI transformation demands patience and pragmatism. The Chronicle operates in the tech epicentre of the world, and that means the company has to learn fast — but also protect what makes journalism different.

At AWS, executives Michael Schmidt and Demian Hess extended that point to infrastructure itself. They described how provenance standards such as C2PA and Model Context Protocol Servers are being built into cloud systems so that news organisations can prove authorship, verify authenticity, and prevent tampering. In a world of synthetic content, they argued, infrastructure is integrity.

Collaboration and intelligence

While the study tour showcased cutting-edge technology, its most repeated theme was human collaboration. AI, tour participants heard again and again, is only as good as the knowledge it organises. From Otter’s “corporate knowledge base” to Stanford’s data-sharing alliances, companies demonstrated how institutional intelligence multiplies when silos break down.

For newsrooms, that means moving beyond tool adoption to workflow redesign: capturing meeting data, codifying editorial reasoning, and making institutional context searchable. The reward isn’t just efficiency; it’s resilience — continuity of knowledge across teams and time zones.

At You.com, CEO Richard Socher described these answer engines as chat-based systems that perform research, summarise sources, and automate verification. His vision echoed the news industry’s need for synthesis rather than speed — technology that amplifies human judgment instead of replacing it.

The leadership challenge

By the study tour’s end, one conclusion united Silicon Valley technologists and visiting publishers alike: AI’s real disruption is managerial, not mechanical. But companies need to find out how and where to bring it in.

Many tools now exist. A differentiator will also be leadership — how to bring along the culture and not what they automate.

INMA participants left with a clearer sense of what that leadership entails:

• Treat AI as infrastructure, not initiative. Embed it in budgets, training, and governance. Create pilot project.
• Build transparency into systems. From data collection to content provenance, traceability is the new credibility.
• Invest in human expertise. AI can summarise, but only journalists can interpret.
• Collaborate. Partnerships with technologists, academics, and peers will define who shapes the next standards.

Rebuilding trust — and the road ahead

What began as a tour of tools ended as a meditation on trust. Across every stop, from Palo Alto labs to San Francisco newsrooms, the message was clear: AI is reshaping not just how journalism operates, but why it exists. Its true potential lies in rebuilding credibility, collaboration, and connection in a fragmented information age.

Participants left the Silicon Valley study tour with a shared sense of urgency — and optimism. Urgency, because the pace of change now demands structural reinvention, not surface-level adoption. Optimism, because the same technologies that once seemed to threaten journalism now hold the power to strengthen it, provided they are used with conscience and transparency.

As INMA’s Tech & AI Study Tour concluded, one truth crystallised: The future of media won’t be built by algorithms alone. It will be shaped by leaders who fuse technical literacy with journalistic values — who see AI not as an existential threat but as an existential test of purpose. The companies that rise to that challenge will not just survive this transformation; they will define what trustworthy media looks like in the AI century.

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