Speakers across INMA’s Media Tech & AI Week in San Francisco made clear last week the economics of AI go far beyond automation. 

AI is transforming how value is created, captured, and distributed across the media ecosystem. Venture capitalist James Cham, a partner at Bloomberg Beta, challenged executives to think of publisher content not as “data” but as intellectual property in a marketplace increasingly mediated by machines.

As AI agents begin to summarise, answer, and recommend, the traditional paths of discovery and monetisation — search, social, programmatic — are being rewritten.

Cham described how AI’s “long tail” dynamics will erode one-to-many publishing economics. In a world of infinite content and hyper-personalised interaction, what once relied on scale and CPMs now depends on intimacy and CPA (cost per action). Creation becomes cheap, distribution smarter, and loyalty more dependent on identity and trust. 

His point: Media companies must stop optimising for mass reach and instead design for relationships, micro-conversions, and verifiable expertise — the elements algorithms will increasingly favour in AI-driven ecosystems.

The broader takeaway echoed throughout the week: Media companies are not competing merely for attention but for machine recognition — ensuring their signals, brands, and rights are visible and valued by the systems that now mediate access to information.

Advertising in the post-platform economy

From the Silicon Valley and San Francisco study tour to the conference stage, AI’s disruption of advertising was seen not as an apocalypse but an opportunity to rebuild value from first principles.

At Cloudflare, executives demonstrated tools that allow publishers to control who scrapes their sites and how often. This included a “pay-per-crawl” beta that could underpin a fairer value exchange between content owners and AI developers.

Rather than being harvested without consent, news publishers can begin to price access to their content — potentially replacing the open, extractive data economy with a transparent, transactional one.

Meanwhile, Vermillio’s TraceID showed how watermarking and monitoring technologies can detect where and how a news publisher’s text, images, or video have been absorbed into AI models. The technology can reverse-engineer AI outputs to reveal how news publisher data is used — making possible real licensing negotiations or royalty frameworks akin to the music industry’s Content ID.

Such traceability, once theoretical, now offers the foundation for new media-AI partnerships built on accountability rather than guesswork.

In sessions with media leaders, speakers argued advertising revenue in this new landscape will depend on quality signals: provenance, performance, and partnership. As one executive summarised, monetisation will shift “from impressions to impact.” 

Native and branded content, powered by first-party data and AI-assisted targeting, may thrive where open programmatic inventory declines. AI will automate ad production, testing, and optimisation — but its real potential lies in aligning campaigns with authenticated content and trusted contexts.

E-commerce and native content find new footing

E-commerce and native advertising are also being redefined by AI’s capacity to personalise, predict, and perform. Study tour discussions pointed to a convergence between editorial storytelling and shoppable experience design.

The thread running through these examples: Context is becoming programmable. News publishers are starting to integrate machine-readable metadata — on products, tone, and trust level — so that AI systems can understand not just what a story says but what it can sell or support.

This represents a new kind of native advertising, in which relevance is determined algorithmically but trust remains human-engineered.

Cham’s caution from the conference resonated here, too: AI will reward those who automate workflows but keep creative differentiation. The organisations that build lightweight, iterative experiments — what Wharton’s Ethan Mollick calls “lab models” — will find faster product-market fit than those waiting for grand strategies.

Rethinking subscriptions and user value

At Microsoft’s MSN, Aparna Lakshmi Ratan explained how large-scale recommendation systems and LLM-powered summarisation (“gems” and Copilot Daily) are driving a more personalised news experience, improving engagement and discovery for quality journalism.

While these features may seem technical, they represent a profound shift in subscription logic: Audiences now expect relevance as a service.

Similarly, LinkedIn positioned itself as a top-of-funnel channel for news discovery, especially as search referral traffic declines. By curating professional perspectives, LinkedIn creates a halo of authority around publisher content — an environment that fosters trust and potential conversion. In this model, brand reputation and verified expertise are the new currency for audience growth and retention.

The INMA sessions underscored how subscription businesses must integrate behavioural data, emotional intelligence, and AI-driven personalisation without losing transparency. If AI systems become the intermediaries of discovery, reputation must be machine-readable — through verifiable bylines, clear sourcing, and brand identifiers that AI can display or interpret accurately.

Speakers noted “super users” — subscribers with deep engagement and advocacy potential — are key to sustainable economics. These users not only pay but participate: sharing, training, and influencing the algorithms that determine future visibility.

Incentivising them through community, customisation, or co-creation may prove more valuable than incremental pricing tactics.

New economic frameworks for the AI age

A recurring theme of the week was that AI requires rethinking the very architecture of value creation. 

The study tour’s visit to CTGT illustrated this with its focus on transparency and auditability in AI systems. Founder Cyril Gorlla argued that just as cybersecurity evolved from compliance to core strategy, media companies must build “trust infrastructure” into their AI use. That means knowing exactly how data is used, measured, and monetised — not only by others but internally across the newsroom and product lines.

In the conference’s closing podcast, The Defrief, conference curator Jodie Hopperton talked with co-moderators Robert Whitehead and Nicki Purcell about how this re-architecture is already visible in news publisher experiments worldwide. 

AI is accelerating both fragmentation and recombination: advertising models merge with subscriptions, e-commerce blends with editorial, and data partnerships redefine the edges between competition and collaboration.

The “middlemen” of the old digital economy — ad networks, social algorithms, and search engines — are being replaced by intelligent agents that operate on behalf of users. The question for media leaders is how to ensure those agents still recognise, represent, and reward professional journalism.

One conclusion shared across sessions: Sustainable business models in the AI era depend on coordination, not competition. As Cham put it, media companies are not in a prisoner’s dilemma with tech giants but a coordination game — where shared standards, content identifiers, and common frameworks can lift the entire ecosystem. Giving away distribution once weakened the industry; giving away data could be worse.

To thrive, news companies must develop the technical literacy, partnerships, and policy influence to define their own terms of engagement — from how data is licensed to how content is presented in AI-driven interfaces. The tools now exist to enforce value; the challenge is using them collectively.

Takeaways for news executives

Across the week’s sessions, five clear imperatives emerged for media leaders:

1. Price and protect data: Treat content as licensable IP, not as free training material. Invest in watermarking, monitoring, and access controls.

2. Design for discoverability: Make brand trust, bylines, and sourcing visible to machines, not just readers.

3. Automate with intent: Identify which workflows can be fully automated to fund creativity elsewhere.

4. Rebuild relationships: Focus on community, participation, and identity-based engagement rather than anonymous reach.

5. Collaborate to compete: Develop shared frameworks for data use, AI ethics, and content standards that strengthen the collective negotiating position of media.

As INMA leaders during The Debrief noted, the coming year will test how well news companies translate these lessons into revenue. Sustainable business models will not come from nostalgia or defensive tactics, but from curiosity, collaboration, and the courage to build new markets for trustworthy information.

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