AI is reshaping advertising, and news publishers must quickly adapt to those changes to thrive.

During the INMA Advertising Town Hall, Gabriel Dorosz, lead of the INMA Advertising Initiative, looked at what changes it has already brought, what’s ahead, and its implications from an advertising perspective.

“This is the most consequential and transformative media tech evolution of our lifetime,” he said. “We’re just at the beginning.”

Dorosz urged attendees to think of AI as both a new Internet and a new OS — an infrastructure-level shift that will redefine how advertising is created, delivered, and measured.

He acknowledged the expected path that will take us from Artificial Intelligence to Artificial General Intelligence (AGI), which is when “AI will potentially behave in a human-like way across all tasks.” The next step would be Artificial Superintelligence (ASI), when AI becomes smarter than humans.

However, Dorosz focused his analysis on the tangible disruptions already underway. “History doesn’t repeat, but it rhymes,” he said, warning that early winners in the AI race may not be the long-term leaders.

He pointed to Apple and Amazon as the most strategically positioned AI players: Apple because it has access to so much of our personal information and Amazon because it owns our shopping data. But, he said, “There are still so many chapters of this story yet to be written.”

Whilst AI is the most transformative technology of our time, in his opinion, it is currently overvalued.

“What we’re seeing right now is this clash between the more optimistic visions and the reality of technical adoption, process integration, and change management,” he said. “There’s also a lot of discussion as to whether or not AI will make advertising unnecessary.”

His answer to that question was a firm no: “Shopping and buying are not the same thing,” he reminded the audience. AI may streamline purchasing, but brands will still need to influence decisions.

He described the current moment as an “impression realignment,” not a collapse, citing estimates that publisher ad exposure is down 30%-40% on average.

Advertising, he argued, remains essential to profitability. Meta, Google, and Amazon all rely heavily on ad revenue, and AI companies will need indirect business models to scale.

“You can’t get to five billion users with a paid offering,” he said, quoting Mark Andreessen, co-founder and partner at Andreessen Horowitz and co-founder of Netscape. “You need an indirect business model. Ads are the obvious one.”

When human insight meets AI

The debate over AI and its ability to eventually generate human insight is one that spills over into every aspect of creativity. From an optimistic viewpoint, AI needs publishers more than publishers need it: “AI doesn’t dream, it doesn’t feel, it doesn’t make art; it aggregates it,” he said.

But that raises another question: Are humans even capable of telling the difference between AI- and human-generated content?

Citing the rise of virtual influencers and AI musicians like Xania Monet, who recently debuted on a Billboard chart, he underscored the idea that AI-generated content is rapidly entering the mainstream. And he pondered what that means for the future of advertising.

A recent study by DVJ Insights compared responses to AI-generated and human-generated ads. When viewers didn’t know which was which, performance was rated equal. When they guessed an ad was AI-made, they rated it lower. However, Dorosz said, “They frequently guessed wrong. People were already struggling to tell the difference in the summer.”

He shared two examples of AI ads in action. The first, a TV ad for Kalshi, was created by one person in two days — and for a budget of just US$2,000. And Coca-Cola’s 2025 holiday campaign was produced in one month instead of one year, using five AI specialists and 70,000 video clips.

Still, Dorosz cautioned that “AI alone doesn’t work yet” when it comes to the production. High-end production companies report AI can’t make micro-adjustments or deliver final image quality at the level of traditional methods, so hybrid workflows are emerging, blending AI with human refinement.

Inside the agency world

AI is triggering dramatic restructuring within the agency world, where brands like GM are bringing strategy in-house, consolidating creative across AI-specialised agencies, and calling it an “AI-driven marketing revolution.” Publicis, meanwhile, is thriving by using AI in pitches — including winning Coca-Cola away from WPP in the United States.

The traditional full-time creative services model is under pressure. Big agencies are morphing into data companies, while smaller ones become service extensions for brands.

In the programmatic space, Dorosz described a coming overhaul. AI is poised to disrupt the middle layers — SSPs, DSPs, trading desks — by enabling direct buyer-seller relationships. He called this “agentic infrastructure development,” pointing to pilots at Yahoo and Muse’s Sigma, and new agentic protocols emerging across the ecosystem:

“You’re starting to see essentially a whole programmatic infrastructure redesign that incorporates agents,” he said. “[It’s] too early to mandate usage of these, too uncertain to predict, but also too significant to ignore.”

Imperatives for 2026

Dorosz closed with his imperatives and guardrails for news companies navigating the AI era:

1. Diversify beyond print and static display.
2. Embrace advertising automation.
3. Build a foundation of signals and engage with advanced measurement.
4. Explore revenue diversification and product innovation.
5. Be louder together; collaboration is key.

“As I've said, this is the most significant moment of change in 40 years of media, technology, and advertising,” Dorosz said. “Iterative evolution is survival. Nothing is certain; things will continue to evolve.”

Traffic and AI are intertwined challenges that will define 2026, he said. News publishers must learn to package and sell their audiences, invest in direct channels like events and branded content, and avoid commoditisation by owning their data.

“AI isn’t magic,” Dorosz concluded. “It requires change management, integration, and testing. Move forward, but stay flexible. Watch what others are doing, and carry forward-looking skepticism as you approach this.”

Photo credit: Adobe Stock Urupong.

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