Advertising in AI is a trust experiment that marketers cannot ignore | MarTech

The most important advertising moment of 2026 took place on the second Sunday of February. It wasn’t the most popular Super Bowl spot or the most cinematic brand song. It was a pointed message embedded in one of the game’s most self-aware ads.

In an effort to position himself against a rival, Claude revealed a line that may be approaching the industry faster than it realizes: What happens when artificial intelligence platforms start making money from advertisers?

Claude’s creator, Anthropic, built its spots around a simple promise: “Ads are coming to AI. But not to Claude.” The humor worked because it dramatized an uncomfortable future: a vulnerable question punctuated by a sneaker pitch, a relationship problem accompanied by a dating app promotion, startup advice followed by a payday loan offer.

The campaign set Claude apart from ChatGPT, at least for now. But the subtext was bigger than the competitiveness. The relationship between humans and AI is evolving.

The ads referenced OpenAI’s decision to test ads in ChatGPT. Not inserted into the comments, as the satire suggested, but shown below the answers for users with a free or entry-level subscription. The placements are labeled ‘sponsored’ and have no influence on the results, according to OpenAI management.

At first glance this seems simple. But AI is not experienced as a billboard or a banner. It is experienced as a conversation. As help. Increasingly as company. That context changes the stakes.

The Facebook echo

The tension was reflected in a recent New York Times op-ed by former OpenAI researcher Zoë Hitzig titled “OpenAI is making the mistakes Facebook made. I’m quitting.”

She recognizes a simple economic truth: AI is expensive to use, and advertising can be a crucial revenue stream. But she warns of something deeper: the ethical tremors that occur when monetization models begin to rely on patterns of human thinking.

We’ve seen this movie before. In its early years, Facebook promised users meaningful control over their data and even the ability to vote on policy changes. Those commitments faded as advertising revenues soared. Financial incentives reshaped the product. The product has reshaped behavior. Confidence dissolved slowly and perhaps imperceptibly.

That’s why, even if OpenAI insists that ads and replies won’t cross streams, the shift itself matters. He has opened the stable door and is leading the horse outside by the reins. Once advertising is in the mud, it tends to buy something. (No pun intended.)

Trust is not just about privacy

Why is this so crucial? Because trust is not just a privacy policy. It is an expectation: the emotional contract users think they are entering into a contract when they type something personal into a machine.

In my book ‘Appreciated Branding’ I argue that brands earn trust when their intentions are clearly aligned with human needs, not when they quietly repurpose those needs as leverage for commerce.

The moment a platform converts empathy-seeking input into ad-adjacency, the emotional math changes. AI advertising exposes a monumental cultural fault line: are AI tools environments for honest assistance or channels for monetization?

In traditional ecosystems such as search engines, social feeds or television, we have a contextual contract. Ads live on the edge. We expect them and divide them into boxes.

But in AI chat, the perimeter disappears. The interface is the conversation, like talking to a therapist who has a part-time job selling comfort animals.

There is no sidebar or separate ad distraction. The experience is immersive and relational. When users feel like their intimate questions are undermining someone else’s income, the safe space becomes polluted – and contagion spreads faster than clarification.

From a valued brand perspective, this is a difficult bell to ring. Remember, trust is not ownership. It is repeatedly reinforced by signals of both attunement and restraint.

Brands that operate with empathetic transparency understand that short-term monetization gains can create long-term relational losses. Once users suspect ulterior motives, they withdraw, not only behaviorally but also emotionally.

Embedding ads into an interface where users share personal concerns risks changing AI’s identity from trusted helper to commercial shill. Trust leaves the chat and something much more expensive than infrastructure breaks.

The business case for restraint

To be clear, the business pressures are real. AI infrastructure is extremely expensive. Free levels need support. Investors expect returns. Advertising is a proven, scalable revenue generation engine.

But here’s the strategic question marketers should be asking: what if AI monetization erodes the trust that makes AI valuable?

If users come to believe that their personal input indirectly drives commerce, they will adapt, self-censor, withhold context, and look for paid alternatives or new platforms that promise neutrality.

In other words, the data source is drying up. Advertising within AI, both in chat and around it, could cause a subtle but devastating change in behavior: less honesty, less vulnerability, less richness of interaction. Ironically, that reduces the effectiveness advertisers hope to achieve.

A different path for brands – and the counterargument

This is where marketers need to think differently. If AI platforms can continue to be an environment where people feel understood without being sold to them, brands have a significant opportunity to gain trust. Make sure they can be found through AI visibility, not paid AI placement.

AI already rewards brand clarity, usability, and problem-solving partnerships that preserve user freedom. That’s the valued brand principle at scale: solve first and then sell as a byproduct of solving.

Platforms that maintain a visible firewall between assistance and monetization may discover something counterintuitive. Maintained trust increases lifetime value. Brands that respect the emotional gravitas of AI interactions can earn deeper loyalty than brands that chase opportunistic impressions.

History complicates this story. I’m old enough to remember consumers saying they would rather walk out naked to get their newspaper than put their credit card number on a website.

We adapt. Standards evolve. What feels invasive today may become common tomorrow. It is entirely possible that clearly labeled, well-regulated ads will become culturally acceptable under AI responses. That users draw their own boundaries and go further. That trust is being recalibrated rather than collapsing.

But the difference here is intimacy. Credit card information was transactional. AI conversations are relational. Once trust is broken, it is not restored as easily as digital payment habits.

The real experiment

Advertising in AI is not necessarily immoral. It may even be economically necessary. But it’s a trust experiment, and trust experiments don’t provide unlimited retries.

If AI platforms make miscalculations and users feel like their vulnerability is quietly being monetized, the damage will extend beyond one company’s quarterly profits. It will reshape expectations of human-technology interactions and shift the cultural agreement from “this tool is here to help me” to “this tool is here to extract value from me.”

Once that agreement changes, it will be far more expensive to rebuild than any data center ever built. For marketers who see this happening, the lesson is bigger than AI. Trust is not a characteristic. It’s infrastructure. Selling the land underneath will be an irreversible one-way transaction.