I have been writing about news media strategy for 36 years. I have sat through more news industry conferences than I can count. I know what it looks like when a field is fragmented and what it looks like when it is finding its footing.

Last week at the INMA World Congress of News Media in Berlin, I watched something I have not seen before: a global news industry quietly agreeing with itself. 

That sounds like a small thing. It’s not. 

For most of the last decade, INMA World Congresses have been arenas of productive disagreement. Should we put up a paywall? Should we chase social? Is video an opportunity or a money pit? Is AI a tool or a threat? Is advertising worth defending or is the future entirely reader-funded?

The disagreements were not symptoms of confusion. They were symptoms of a news industry still working out its diagnosis. You cannot agree on a treatment when you cannot agree on the disease.

What changed in Berlin is the diagnosis. Across 70+ hours of programming from a two-day conference, three study tours, and six seminars, the record 750+ media executives attending INMA Berlin landed on a shared analysis with a tightness I would not have predicted 12 months ago. 

Three keynotes set the frame, and they converged tightly. 

INMA President Gert Ysebaert opened the Congress with the pivot from reach to being chosen. Distribution power has moved to platforms and AI agents, and the era when being found was enough is over. The legacy moats — distribution, scale, breaking-news speed — are gone. AI is collapsing the marginal cost of content toward zero.

The only path forward, Gert argued, is solidarity as collective leverage against platforms that have grown too large to negotiate with individually.

My own opening keynote argued the only defensible moat left is brand as an operating system: provenance, judgment, and institutional identity encoded into every touchpoint, the one asset no platform can take and no AI can replicate.

Brand is not a marketing layer. It is the operating logic of a publisher’s relationship with readers, advertisers, AI agents, and regulators. Everything else — the AI workflows, the subscription funnels, the licensing terms, the platform sovereignty work — is downstream of whether the brand operates coherently across every surface.

Julia Becker, Funke chairwoman, closed the loop on the editorial side. Journalism’s unique product is trust. That trust is now under existential pressure from algorithm-optimised engagement and synthetic content. The strategic response cannot be to chase speed or sensation. It must be to double down on credibility, relevance, and local proximity — and to stop underselling our work. Quality journalism is a premium product, not a commodity.

Platforms have eaten distribution. AI is commoditising the artifact. The only path through is identity, trust, and direct relationship, operationalised as brand across every touchpoint and defended collectively through institutional infrastructure.

Identity. Trust. Habit.

Kristin Heitmann’s framing of the Associated Press’ strategic pivot put it most precisely: “We are no longer just in the news business; we are in the trust business, and that trust is now a quantifiable, licensable asset.”

That sentence is the diagnosis INMA members arrived at in Berlin, expressed by the cooperative news agency with one of the longest institutional histories in our membership. It is what every other session in the week was, in its own way, working out.

That is the diagnosis. What surprised me in Berlin is that INMA members have begun to converge on the treatment as well.

What we agreed on

Four shifts emerged in nearly every room, from very different people in very different markets. They are not competing strategies. They are pieces of the same operating model. 

The first is that AI is no longer just a trend. It is forcing deep operating model transformation. 

The most aggressive language of the week — “embrace AI or die,” “use AI daily or step aside,” 97% of code now written by agents — came from the largest, most established legacy players, not from startups. Axel Springer, Reuters, Hindustan Times, BBC, Tamedia, Mediacorp, Funke, Ippen, Bauer, Holtzbrinck, FT, Guardian, Spiegel, DPA, RND, Tagesspiegel — many of them presented an AI-native operating model and nearly all showed AI moving from experiment to infrastructure.

INMA Product & Tech Initiative Lead Jodie Hopperton’s Agentic AI Seminar and Newsroom Innovation Initiative Lead Sonali Verma’s AI in the Newsroom Seminar introduced a new vocabulary — skills over agents, conductor orchestration, AI super-user cohorts, “AI architect,” and “journalistic creator” as the two emerging journalism roles — and the room absorbed it as common sense rather than novelty. 

One study tour briefing reported 96% of German publishers are already using AI in their newsrooms. Hindustan Times reported 60%-70% engineering time savings and a software development cycle compressed from three or four months to weeks. OpenAI’s cost curve, presented as background context rather than as headline news, showed input token costs dropping from US$60 per million tokens in 2023 to US$0.05 today. 

Two years ago, the news industry’s question was whether to adopt. Now the question is which agentic workflow patterns scale, how to measure adoption depth, how to govern autonomy without becoming the bottleneck.

Stibo’s reasoning note put it sharply: If every AI action requires human review, the review becomes the bottleneck, and the efficiency gains evaporate. The news industry is now wrestling with that problem, not the previous one. 

The strategic question underneath all of this came from Sonali, who opened the AI in the Newsroom Seminar with what is now the news industry’s working operational test: “Why would anyone come to me for information rather than going to ChatGPT?”

Every AI workflow, every newsroom adoption decision, every product change should answer that question or it does not justify itself. Sonali’s diagnosis is that AI-mediated information is producing sameness and flatness — what I called a "swamp of sameness” in my keynote — content that converges on the median because the models are trained on the median.

The publisher’s job is to be the answer to her question: to produce, on every surface, what an AI cannot. The human-visibility tactic she advocates — making editors and reporters visible through events, videos, and bylines that mean something — is the simplest and most replicable single move newsrooms can make to demonstrate that real people, with judgment, are doing the work.

The Associated Press operationalised the same insight on the opposite surface. Where Sonali’s tactic is to make humans visible to human readers, Kristin Heitmann’s strategy at AP is to make AP’s journalism legible to machine readers.

Both moves treat the audience question seriously. Both answer it on their own terms. AP’s pivot is from selling journalism to humans to selling structured, machine-readable intelligence to AI models, risk platforms, and trading algorithms.

The OpenAI deal three years ago was the pilot. The Snowflake data-marketplace distribution and the new verification services are the build-out. The strategic move is to treat the audience for AP’s journalism as including machines — not as an inevitability to defend against but as a customer to serve, on AP’s pricing terms, with AP’s editorial standards encoded in the metadata.

The four pillars Kristin laid out — B2B media licensing, direct-to-consumer, election data (counted since 1848), and what she called “Intelligence” as the new frontier — are the most fully articulated answer in the news industry to the question of where the AI-native news business is heading.

Most publishers have not yet asked the question. AP has built a P&L around the answer.

The second is that reader revenue is the center of gravity, and the playbook is maturing. 

Greg Piechota’s subscription and financial benchmarks tell the structural story. Roughly 60% of publisher revenue is now reader, 29% advertiser, 11% other. Some 22% of benchmarked business units already earn higher digital revenues than their newsroom costs.

The Subscription Growth Seminar at the end of the week in Berlin layered five different angles: Mediahuis on brand differentiation as a funnel KPI, Tagesspiegel on editorial-led product reorganization, PD Digital on data and tech stack modernisation, Bonnier on the “+Allt” super-bundle, The Washington Post on flexible Week, Day, and Article passes — and they fit together rather than competing.

The biggest single finding inside the subscription discipline this year was that growth is driven by volume of users at the top of the funnel — not by paywall tightness, operational efficiency, or pricing optimisation. Fast growers had more users getting value, hitting paywalls, starting checkouts, and buying. Reach and loyalty turn out to be empirically correlated, not opposed.  

The maxim that has guided strategy for a decade — “we are in the business of loyalty, not reach” — is empirically wrong. Bonnier’s funnel made it explicit: Registration to login to first-party data to subscription is one continuous pathway, with logged-in users generating a 100% advertising price uplift and subscribers generating four times the ad revenue and four times the pageviews of non-subscribers.  

Reader revenue and advertising revenue are not parallel businesses. They are the same business, with the same engine, sold twice. 

The super-user finding compounds this. Only 1% of users generate 26% of pageviews on average across 300 brands and up to 81% of total revenue at one European publisher. The most engaged readers are simultaneously the most valuable subscription targets and the most valuable advertising targets.  

The German market made the parallel point at the premium end: Tagesspiegel Background at €180 a month with near-zero churn, Table Media at 1.5 million total subscriptions across 13 verticals, SZ Dossier at €1,000 per year, Politico Pro at US$7,000-US$12,000 annually, and Media Pioneer at a €10,000 access tier.

The B2B briefing layer over consumer news is no longer an experiment. It is a category. 

The third is that mobile, vertical, video, and native are no longer optional. 

Nobody defended text-only anymore. 

The economic forcing function is the 5-to-1 revenue ratio between vertical video and display advertising.  

The behavioural forcing function is the Reuters Institute data Kerstin Hasse, lead of the INMA Young Audiences Initiative, cited from her Young Audiences Seminar: 39% of young people use social media as primary news source, 51% pay more attention to creators than to news brands, 64% of under-25s consume news daily. Die Zeit’s TikTok team at 96 million January views. Mediacorp’s 2.7 billion TikTok views, with 350% growth attributed to AI enablement. Sky News at 11 million TikTok followers and 85% year-over-year video view growth. Bild at 70% mobile traffic, high 80s on weekends. Spiegel converting to vertical video despite acknowledging it looks bad on desktop because 80% of the audience is on mobile, and the format works there. 

Kerstin’s framework was the most insistent: The funnel is dead, replaced by a Discovery → Trust → Loyalty cycle that runs on platforms publishers don’t own. Her five principles — discovery happens on platforms, creators over legacy title, authenticity builds trust, community and belonging, diversify formats — are now the operating principles of the youth strategy, full stop. 

The creator question moved from “should we?” to “build, partner, or train?” The case made by Bild’s Paul Ronzheimer’s case — 250,000 to 350,000 daily podcast listeners with 95% retention through 40- to 45-minute episodes — illustrated that the personal-brand journalist works inside a media house, too, but only with infrastructure, security backing, and a deliberate trust posture. 

The fourth is the shift I think the post-Congress coverage will most underplay: advertising’s quiet refusal to be written off. 

INMA benchmarks say 29% of publisher revenue still comes from advertisers. That is roughly the same size as subscription revenue two decades ago.

The news industry has spent so much of the last five years telling itself a reader-revenue story that we have started talking as if advertising were already gone. It is not gone. It is the under-weighted second pillar, and INMA Berlin pushed back on the residual treatment harder than I credited at the time. 

The reframe INMA’s Gabriel Dorosz, lead of our Advertising Initiative, made on the main stage was structural. The global advertising market is roughly US$1 trillion today, heading to US$1.3 trillion. Digital is 80% of that and rising toward 85%. News publishers collectively receive only 3%-5% of the total, and that share is declining. Google, Meta, and Amazon capture 60%-70%, and their share is rising. 

The question Gabe asked — and I think it is the question of the next 12 months — is not whether the news industry can defend its slice. It is whether the industry can build the standards, formats, and collective position to take more of the US$1 trillion it currently does not get. For two decades the publisher advertising conversation has been about sharing a shrinking pie. 

INMA Berlin asked, for the first time I can remember, whether that is even the right pie. 

The Advertising Seminar made the case concretely:

• Bonnier’s logged-in funnel produces a 100% price uplift on first-party-data ad inventory and four times the ad revenue per subscriber.
• News Corp Australia’s Engaged Reach metric — pageviews divided by unique audience — reorganised editorial around monetisable engagement, drove a 600% lift in app downloads and an 800% lift in ad yield, and quietly demonstrated that a clean KPI shared across editorial and commercial is itself a strategy.
• Mediahuis pivoted from selling reach to selling impact, trust, and consultative service with measurable client wins.
• Styria’s COPE in Austria built a first-party data alliance to claw back the 86% of digital ad spend captured by global platforms.
• Bonnier in Sweden entered in-store retail media through visual art and, for the first time in 25 years, had FMCG clients like Procter & Gamble calling them.
• Russmedia demonstrated outcome-based “utility inventory” ads — working tools embedded in editorial, priced per use, built in 14 hours by sales teams using vibe coding.
• The IAB’s open-source toolkit moves the entire monetisation stack to the CDN edge, recapturing the 30%-35% CPM loss Apple’s privacy moves inflicted and bypassing most ad blockers. 

The super-user insight cuts both ways here, and INMA Berlin’s advertising track was the room that named it. The most engaged readers are the most valuable to advertisers. One Latin American publisher reported 1% of subscribers generating 72% of combined subscription and advertising revenue. The New York Times previously reported that its most engaged users are the most receptive to advertising.

The “horrible ads turned off for subscribers” pitch we have spent a decade making is not just bad messaging, it’s bad strategy. It is throwing away the highest-yield advertising inventory the publisher owns to the readers most likely to engage with it. The flywheel between reader engagement and advertising performance was inverted by digital display’s bad incentives. It can be inverted back, and The New York Times’ fifth straight quarter of double-digit digital ad growth is the proof that it pays. 

And then there is the US$2.8 billion the news industry is losing annually to brand-safety technology that classifies professional news as unsafe and Meta as safe. A tennis story blocked for the word “smash.” A vineyard story blocked for “alcohol.” Vendors paid each time they block, creating a perverse incentive to over-block.

The technology to fix it — context-aware, AI-powered alternatives — exists. The blocker is collective will, not technical capability. This is exactly the SPUR pattern applied to advertising, and it is exactly the playbook the AI track ran. It works in advertising too, if the membership chooses to run it. 

Advertising is not the past of news media revenue. It is the under-weighted second pillar of its future. 

What the convergence looks like in practice 

The thing that makes me confident this is real, and not just a well-curated World Congress agenda, is the same patterns appeared independently across all three study tours. 

The study tours visited Tagesspiegel, Axel Springer, RND/Madsack, Table Media, Der Spiegel, Die Zeit, NZZ, DPA, The Pioneer, MorningCrunch, Bauer, FAZ, SZ, taz, Ippen, Funke, and Correctiv. Different organisations, different scales, different business models, different ages and editorial traditions. The patterns were the same. 

Editorial collaborating with product. Cross-functional teams replacing silos. AI integrated into the CMS with human-in-the-loop quality gates. Vertical video as core editorial output. Direct-traffic share as the defensive moat. The B2B briefing as the high-margin layer. The personal-brand journalist as the new acquisition channel. First-party data as the foundation for both subscription and advertising. The platform as a frenemy to be negotiated with from a position of standards. 

This is what convergence looks like. Not everyone doing the same thing, but everyone solving the same problem with overlapping vocabulary, comparable architectures, and increasingly compatible playbooks.

The closest analogue I can think of is the moment in the mid-1990s when newspaper executives stopped arguing about whether to put their content on the Web and started arguing about how. The argument about how is much more productive than the argument about whether. It produces benchmarks. It produces standards. 

How INMA is built to act on this 

If the diagnosis has converged and the treatment is becoming visible, the next question is institutional. Does INMA’s operating structure match the news industry’s strategic posture?

Looking back at the six seminars that closed Berlin, I think the answer is “yes” — and I want to make that visible, because the seminars are not six separate tracks. They are the six initiatives through which INMA operationalises the response to everything in this post. 

Greg Piechota’s Readers First Initiative operationalises the reader-revenue shift. The Subscription Growth Seminar’s five-case spine — Mediahuis, Tagesspiegel, PD Digital, Bonnier, Washington Post — is the working playbook the membership now has access to. The Oslo Subscription Summit in March 2027, the benchmark services Piechota leads, and the new report published the week before the Berlin Congress are the instruments.

The super-user finding shared with advertising, the funnels to relationships insight, and the reach-loyalty correlation are all Readers First findings now embedded in the news industry’s vocabulary.

The Advertising Initiative, relaunched a year ago, operationalises the second-pillar shift. Gabe Dorosz’s plenary and the Advertising Seminar are the public face of work that INMA members has been waiting for. The trillion-dollar reframe, the first-party data architecture, the brand-safety collective action, the engaged-reach metric — these are not isolated insights.

They are the substance of an initiative that needs the same narrative gravity Readers First has built over the last eight years. The membership has the substance, as exemplified by the representation from Mediahuis, Bonnier, COPE, Russmedia, and News Corp Australia. The Advertising Initiative is how we give it institutional shape — now punctuated by a Media Advertising Summit & Study Tour in New York in December.  

Jodie Hopperton’s Product & Tech Initiative, with the agentic AI track as its leading edge, operationalises the AI shift on the product and platform side. The seminars with OpenAI, the Multi-Agent AI for Publishing session, the Holtzbrinck AI Hub case, the Media Tech & AI Week in San Francisco — these are the architecture conversations that turn AI from a tool into an operating system.

The shift from agent-as-novelty to agent-as-infrastructure is the work this initiative is running.

Sonali Verma’s Newsroom Innovation Initiative operationalises the AI shift on the editorial side. The advisory council of global editorial leaders, the four reports already produced, the bi-monthly newsletter, the BBC and Tamedia and Mediacorp and Stibo cases the AI in the Newsroom Seminar surfaced — these are how editorial leaders learn from each other about adoption depth, governance, super-user cohorts, and the human-in-the-loop discipline that prevents the review-as-bottleneck failure mode.

The “why would anyone come to me instead of ChatGPT?” question she poses is the news industry’s working operational test.

Kerstin Hasse’s Young Audiences Initiative, launched in January, operationalises the format-and-creator shift. Hasse’s five-principle framework is the most cleanly articulated operating model produced by any of the initiatives this year.

The September 2026 Young Audiences Summit, the No One’s Reading This podcast, the Ask Me Anything sessions, and the case work with Spot on, The Economist, Podme, Grupo Clarín, and the rest — these are the channels through which the news industry will move from talking about reaching younger audiences to operationally reaching them. 

Robert Whitehead’s Digital Platforms Initiative runs underneath all of the above. SPUR, RSL and the RSL Collective, the CDN partner work with Cloudflare, Akamai, and Fastly, the bot-class separation standards, the licensing-first regulatory engagement — these are the sovereignty layer: 21 public media-AI deals signed, 52 active lawsuits with no decisive precedent yet, 1,500 publishers engaged in the RSL framework.

This is not one shift among four. It is the foundational infrastructure that determines whether the other four shifts have a stable market to operate in. Gert Ysebaert’s opening call for solidarity as leverage is, operationally, the Digital Platforms Initiative’s mandate. 

The six initiatives are how the diagnosis becomes action. The World Congress architecture and the initiative architecture are the same architecture, and Berlin made that visible in a way previous Congresses have not.

The argument for building the marketplace

If SPUR, RSL, and RSLC are the infrastructure, the question becomes whether publishers will actually use them — and to what end. 

Two Berlin sessions answered that question in complementary angles: Jonathan Roberts of People Inc. made the strategic case for why publishers should build the licensed content marketplace at all. Caitlin Clark of the Financial Times showed what executing that posture looks like inside a single publishing house.

Together, they moved the conversation from capability to conviction: The tools exist, the market is forming in real time, and the publishers who treat licensed content as a business rather than a grievance are the ones already negotiating from strength.

Jonathan gave that infrastructure its strategic argument. His People Inc. session was the clearest articulation anyone made of the publisher posture for the AI economy: drivers, not passengers.

AI needs three pillars — chips, power, and information — and information is the under-funded leg. The current paradigm of free scraping produces degrading inputs that produce degrading outputs and is unsustainable on its own terms. Publishers are best positioned to build the licensed, structured, rights-cleared content marketplace that enterprises and AI companies actually need.

The market data he laid out is real: Roughly 30 venture-backed companies in “content as a service” in the last year, hundred-million-dollar funding rounds being raised, a million crawls roughly priced at US$10,000-US$50,000, AI query volume expanding 10-30 times above current human search volumes. People Inc.’s own block-then-license sequence — slowing crawler exfiltration from 100 milliseconds to five to eight seconds, then negotiating from a position of standards — is the operational proof that the SPUR/RSL/RSLC infrastructure can be executed today, not deferred to a policy timeline.

Jon’s line is the one to hold: If publishers do the work, they build a new economy. If they don’t, others will build it in ways worse for publishers and society.

Publishers possess both the strongest incentive and the strongest expertise to make the new system work. The INMA Digital Platforms Initiative is how the membership pools that incentive and expertise.

Caitlin’s session gave the same posture with its institutional architecture. The Financial Times is executing what she called a three-pillar AI engagement strategy:

1. Internal adoption: Editorial augmentation, AI-powered reader features on FT.com, computational journalism in the data team.
2. Policy and standards engagement: Including participation in SPUR and ongoing work with regulators.
3. Structured licensing deals with tech platforms.

The licensing terms Caitlin described are the most fully articulated I have seen from any publisher with required attribution and direct links back to FT.com, AI-generated outputs structured to function like search results rather than substitutes (preserving full-value monetisation on the publisher’s own property), and explicit exit clauses preserving the FT’s ability to walk away.

The strategic bet underneath the architecture is that AI models depending on real-time factual information will, over time, depend on reliable sources — and that dependency is the publisher's leverage. 

Caitlin’s framework is the operational complement to Jonathan’s strategic posture. He says the market is real and growing. She shows how a single publisher engages with it across all three dimensions simultaneously without giving away the business.

What I am still watching

I would not be doing my job if I did not flag what I think the convergence is missing.

Guardian Editor-in-Chief Katharine Viner gave the cleanest counter-narrative of the week. While the rest of the news industry optimised paywalls, flexible passes, and bundle pricing, Viner called paywalls “as divisive as physical barriers” and the contributor model “a political act.”

Seventy-five percent of the world’s population now lives in autocracies, she noted, for the first time since the Berlin Wall fell. She has a point the rest of the room mostly did not engage with. If our heaviest bet is on subscription revenue, are we optimising for the journalism that democracies require or for the journalism the wealthy can afford?

That question is not going to answer itself, and it is one only an organisation like INMA, spanning more than 100 countries and every business model, can credibly convene.

Paul Bascobert at Reuters gave the second counter-narrative. While much of the room is diversifying into adjacent businesses, Reuters is sharpening focus. One thing done well, sold into many windows. The “winemaker not restaurant” framing is a direct critique of the diversification strategies several publishers presented in the same week. 

Both Viner and Bascobert are saying, in different ways, that ownership structure and strategic focus determine which business models are even available. The convergence I described above tends to obscure that. 

A handful of other moments cut against the dominant story in ways worth sitting with:

• Bauer shutting its digital unit.
• Mediengruppe Bayern cash-rich and acquisitive on print.
• Frank Mahlberg at Axel Springer explicitly conceding that digital EBITDA will not replace print EBITDA one-for-one, which is why the affiliate-commerce third pillar is strategic — and there is an ongoing search for more pillars.
• Wikipedia founder Jimmy Wales reporting 8% year-on-year decline in human Wikipedia visitors with a corresponding decline in contributors — an existential funnel problem for the platform that defines trusted human knowledge.
• The Atlantic’s Anne Applebaum, with the historian’s distance, reminding the room that the autocrat’s playbook for capturing media is well-developed and globally portable, and that financial precarity is the lever that activates it.

Wales’ Wikipedia data and Heitmann’s AP strategy are the same finding read two different ways: Wales sees the human reader funnel collapsing as agents intercept readers before they arrive. Heitmann sees the same collapse and treats it as the market opening.

If trust is a quantifiable, licensable asset, then the audience to license it to is precisely the agents that are draining Wales’ funnel. The news industry has spent three years framing AI as a threat to be defended against. The strongest sessions in Berlin — AP, People Inc., Schibsted’s OpenAI deal — framed it as a market to be priced into. That is the more useful posture, and news media companies should adopt it faster than it has been.

Two things were also under-weighted given how much they will matter:

• Audio: Despite Die Zeit’s podcast dominance, Axel Springer’s hourly audio briefing, the BBC’s My Club Daily, and the rise of video podcasting as the bridge format — did not get a forum of its own.
• Trust measurement: Given that every keynote and most plenaries declared trust the product — was almost entirely absent. Everyone said it. Almost nobody showed how they measure whether their own brand’s trust is moving. 

I will leave these as open questions for the year ahead because that is what they are.

What this means for the next 12 months 

I have been thinking about what to take from Berlin as we move toward the INMA World Congress in New York April 26-30, 2027.

The first is that the six initiatives need to operate as a portfolio, not as six tracks. Readers First and Advertising share the super-user finding and the first-party data architecture. Newsroom Innovation and Product & Tech share the AI adoption stack and the human-in-the-loop discipline. Young Audiences shares the creator-economy work with Newsroom Innovation and the format-economics work with Advertising. Digital Platforms runs underneath all five. 

The members who get the most out of INMA are the ones who recognise that the initiatives are connected. Our job over the next 12 months is to make those connections more visible and more operational.

The second is that benchmarking needs to be formalised across initiatives. Greg Piechota’s subscription synthesis is the model. We need some form of this on advertising, on agentic AI adoption depth, on newsroom AI integration patterns, on young-audience reach and conversion, and on platform-licensing economics. The data is now available across enough markets to produce shared scorecards rather than aspirational frameworks.

The third is that the policy voice the membership has asked us to elevate has a clear immediate agenda. SPUR. The RSL Collective. Bot-class separation. Telemetry standards. The US$2.8 billion brand-safety problem. Licensing-first regulatory engagement in the EU, the UK, Australia, and increasingly the United States. Anne Applebaum’s reminder that platform power and regulatory capture are tools authoritarians use should sharpen the urgency of this work, not just for industry economics but for democratic infrastructure.

The fourth is the one I want to close on: operationalisation vs. disintermediation. We spent five days in Berlin describing the moats we still have. Verified human authorship. Editorial judgment. Institutional accountability. Direct relationship. Trust. These are not abstractions. They are operating commitments. The news industry is converging on a shared playbook for defending them — not perfectly, not uniformly, but visibly. The six initiatives are how we move from playbook to practice. The next 12 months will be about whether the operationalisation happens fast enough to outrun the disintermediation.

INMA Berlin made me more optimistic than I expected to feel. Not because the problems are easier than they were a year ago. They are not. But because for the first time in a long time, the news industry is no longer arguing about what the problem is.

Now we have to do the work.

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