News media companies are increasingly comfortable with generative AI, integrating it across departments beyond the newsroom and leveraging it for more than workflows and analytics.

During this week’s Webinar, INMA members were given a closer look at how AI is transforming advertising and what it potentially means for the future.

Sonali Verma, INMA’s former Generative AI Initiative lead and current lead of the INMA Newsroom Innovation Initiative, moderated How to use GenAI for more effective advertising. The Webinar is the fifth in a series with INMA and OpenAI designed to explore how AI can be used within news media companies.

Verma began by asking attendees how they are using AI in advertising.

Based on a poll of participants, the primary uses among news media companies are copywriting, audience analysis, and creative generation.

That’s consistent with what Gabriel Dorosz, lead of INMA’s Advertising Initiative, has seen in companies across the board. He noted that although many publishers are experimenting with small, efficiency‑focused pilots, the broader advertising ecosystem is undergoing a profound structural shift:

“A lot of these examples that I see are very discreet, very tight, very efficiency oriented, integrating into existing processes,” he said. More ambitious, AI‑centric systems are emerging but require longer timelines due to security, technology, and organisational implications, he added.

Rapid expansion

AI‑powered advertising is expanding at a pace that has outstripped earlier forecasts, Dorosz  said. Citing a GroupM study, he noted that global AI‑powered ad spending is projected to reach US$142 billion by 2030, representing roughly 15% of the total advertising market.

More striking, he said, was the finding that “more than 90% of advertising will be AI-enabled in some way by 2029” — a milestone arriving three years earlier than previously expected. Already, by 2025, the majority of ad revenue was touched by AI, with generative tools most commonly used in media planning, activation, and video ad creation.

One significant development for media companies to be aware of is the transformation of large language models into media platforms. He pointed to a recent watershed moment: ChatGPT began serving ads on February 9, 2026, entering the market with a premium CPM of US$60 and a minimum spend of US$200,000. “I point out that that is more on the high side,” Dorosz said.

However, the steep price tag didn’t stop major agencies, including Omnicom, from “diving in head first,” with more than 30 brands joining in the pilot. Meanwhile, Google signalled that Gemini would eventually carry ads, Meta began monetising chatbot conversations, and perplexity abandoned its advertising plans altogether.

The competitive landscape, he said, is “very, very dynamic” and evolving in real time.

Such shifts are reshaping the consumer purchase journey.

Thanks to AI‑driven discovery tools, things that once took weeks of research can now be done in seconds. AI has automated the middle funnel by providing product reviews and comparisons before audiences ever reach a brand’s Web site, he said.

Once a consumer expresses interest, LLMs can instantly narrow options and deliver recommendations, pushing advertising influence toward two extremes: impulse purchases and long‑term, trust‑based decisions such as financial services or automotive buying.

That means traditional retargeting, which has historically been successful for advertisers, is losing effectiveness and instead reinforcing relevance rather than driving decisions.

Where do AI ads belong?

The industry is also seeing a public debate among AI companies about whether advertising should be included in conversational AI.

OpenAI’s move to introduce ads prompted Anthropic to position its chat solution, Claude, as an ad‑free alternative — even using its Super Bowl buy to emphasise that stance. OpenAI countered that advertising makes AI accessible to non‑paying users.

Dorosz said this tension — trust versus monetisation — will shape the competitive landscape for years: “There are so many implications for publishers in all of that, so we all should be paying close attention.”

Creativity remains “an existential debate,” Dorosz said. While studies show consumers often cannot distinguish between AI‑generated and human‑made ads, high‑end creative still exposes AI’s limitations. Dorosz noted that AI struggles with the micro‑adjustments required for premium production, and brands rarely request fully AI‑generated ads.

The 2026 Super Bowl underscored this: AI‑themed ads performed poorly, scoring low for attention, likability, and watchability. The single fully AI‑generated spot — for Svedka vodka — became a “cautionary tale” after “very, very negative” audience reactions. Human‑made, emotionally driven ads — such as the Budweiser spot — dominated consumer rankings.

The most transformative disruption Dorosz sees is unfolding in programmatic advertising. He described AI as an “event horizon” for the buy‑sell ecosystem, predicting that AI will collapse the layers of intermediaries — SSPs, DSPs, and trading desks — and enable direct, real‑time transactions between publishers and advertisers.

CES 2026 marked a surge of agentic AI announcements, with platforms, agencies, and broadcasters unveiling autonomous campaign pilots. A standards battle is now underway between the IAB Tech Lab’s agentic road map and the competing Ad Context Protocol, a contest that will determine how the next generation of AI‑driven advertising operates.

Dorosz offered clear guidance for publishers: prepare data strategies now, but wait for demand and standards before operationalising agentic trading. First‑party data, contextual signals, and advanced measurement capabilities will be essential to participate in future AI‑driven markets.

“Without first‑party data … you risk not being able to tap into demand when it comes,” he warned. “Pay very, very close attention to what’s happening there so you’re in position to act when it’s time.”

AI in action at Jagran

Atul Tyagi, associate vice president and head of audience at the Adtech and Analytics Hub at Jagran New Media, shared insights into how Indian publishers are beginning to deploy GenAI as a practical tool to reshape advertiser relationships and enhance audience intelligence.

Jagran New Media operates one of India’s largest Hindi‑language news platforms, jagran.com, supported by a constellation of niche verticals including HerZindagi (women’s lifestyle), Jagran Josh (education and careers), and OnlyMyHealth (health and wellness).

This diverse portfolio, combined with multilingual content across Hindi, English, and Gujarati, gives the company a broad and highly segmented audience base. That diversity, Tyagi said, created the perfect environment to test AI‑powered, hyper‑contextual brand interactions.

“We realised that through generative AI, we don’t have to send the same messaging to each of our users to extract an opinion from them or get to know what they think about the brand that we want to talk about,” Tyagi said. “We can actually share with a very user-specific group-specific messaging using GenAI and then collect responses and then ask subsequent questions — which is what we have seen that GenAI is very good at: They learn context, and then you can keep on asking subsequent questions.”

Tyagi shared some of Jagran’s early experiments in AI‑driven brand engagement and how they enabled deeper, more personalised insights into consumer sentiment.

In the first example, Jagran conducted a survey among automotive advertisers seeking better demographic insights. Traditional surveys, Tyagi noted, send the same question to every user. But GenAI enabled Jagran to tailor questions dynamically based on user behaviour, geography, past responses, and the type of content a user was reading at that moment.

Instead of presenting a single question about car‑buying intent, the team used GenAI to create multiple variations of the same query, each phrased differently and targeted to specific user segments. If a user indicated interest in buying a car within 12 months, the system automatically generated follow‑up questions about preferred vehicle type — SUV, sedan, hatchback — and then drilled down into brand affinity.

“We designed a set of questions through which we could determine if the user is looking to buy a car and of which brand,” Tyagi explained.

The results provided the advertiser with actionable insights: “For example, we found that car buyers who mostly favoured a Maruti Suzuki car would not be very keen on buying a SUV from them.”

 These insights also enabled Jagran to retarget users for extended brand engagement, allowing advertisers to run multi‑month campaigns that could evolve based on user responses.

In his second case study, Tyagi shared how a major media agency wanted to understand how Indian audiences consume free‑to‑air (FTA) television channels on direct‑to‑home (DTH) platforms. Again, Jagran used GenAI to create multi‑part surveys that adapted to user behaviour over time.

Responses were collected across different viewing segments and completion rates were high: “If you present the survey in their context, they’re very likely to fill it up,” Tyagi said.

The AI‑driven approach allowed Jagran to segment users by age, gender, geography, device type and content affinity, producing granular audience profiles that the agency could use for planning and measurement. More importantly, it demonstrated how publishers can transform surveys from static, one‑off exercises into ongoing, cyclical brand conversations.

AI‑powered audience intelligence is not just a data exercise, Tyagi said. It is a strategic shift. By enabling brands to “engage with the audience in a longer term” and adapt messaging based on evolving sentiment, publishers can create new value propositions that extend far beyond traditional ad buys.

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