AI Advertising: What Happens When Your Assistant Starts Selling to You?

I have a complicated relationship with advertising.

I understand why it exists. Media has to make money. Platforms have to make money. Businesses need customers, and customers need ways to discover businesses.

None of that is particularly shocking.

What gets interesting is what happens when the advertisement stops looking like an advertisement.

That question feels especially important as more people start using artificial intelligence the way we once used search engines.

We ask for recommendations.

We compare products.

We research hotels.

We plan trips.

We ask what to buy, what to eat, where to go, and which option might fit our lives best.

Now advertising is entering that conversation too.

Apparently, We Are Already Asking AI What to Buy

A new report from Adblock Plus looks at advertising inside conversational AI and how consumers feel about it.

One statistic immediately caught my attention.

According to GWI data cited in the report, 49% of internet users had used AI to find information during the previous month. Even more interesting, 23% said AI had become their main source when actively looking for information about brands, products, or services.

Among people using AI for purchase decisions, 33% used it for detailed product information. Another 31% used it for price comparisons, while 30% used it to compare product features.

In other words, AI is not simply helping people write emails and summarize documents.

It is entering the shopping process.

That creates an obvious business opportunity.

It also creates a trust problem.

A Search Result and a Conversation Do Not Feel the Same

I know when I see an advertisement sitting above Google search results.

I may dislike it.

I may scroll right past it.

Still, the commercial relationship is familiar.

A company paid for visibility.

A chatbot feels different.

Conversation carries an assumption of responsiveness. I ask something specific, and the system replies directly to me.

That exchange can feel far more personal than scanning a page full of links.

So when advertising enters the conversation, the boundary between advice and promotion becomes more important.

If I ask an AI assistant which suitcase is best for a six-week trip, I want to understand whether a recommendation comes from the answer itself, an advertiser, or some mixture of commercial and informational systems operating around it.

The distinction should not require detective work.

Seventy-Two Percent Is a Pretty Loud Expression of Concern

The Adblock Plus report commissioned a Harris Poll survey of 2,066 U.S. adults in May 2026.

According to those results, 72% said they were concerned about AI assistants using conversation data to target them with advertising.

Forty-four percent described themselves as very or extremely concerned.

Those numbers make sense to me because conversations with AI can contain information people would never type into an ordinary product search.

Think about the difference.

A search might say:

“comfortable walking shoes.”

A conversation might say:

“I am traveling alone for three weeks, my knees hurt after long walking days, I have a limited budget, and I need shoes that will work with the clothes I am packing.”

That second interaction contains much more context.

Useful context can produce a better answer.

It can also become incredibly valuable commercial information.

The More Helpful AI Becomes, the More Valuable the Conversation Becomes

This is the tension I keep coming back to.

We want AI systems to understand context.

Context is what makes them useful.

If I tell an assistant my budget, destination, preferences, dietary restrictions, and schedule, I want the answer to reflect those details.

But detailed context also tells a commercial system exactly what kind of consumer I am at that moment.

Traditional advertising has spent decades trying to infer intent.

A person searches “flights to Mexico City,” and advertisers reasonably assume that person may be planning a trip.

Conversational AI can potentially receive the intent directly.

The consumer explains what she wants in complete sentences.

That changes the value of the interaction.

The Advertisement May Eventually Know the Conversation Too

Researchers cited in the report are especially interested in what happens if advertising moves beyond clearly labeled boxes.

Dr. Florian Schaub of the University of Michigan School of Information described a future in which AI could generate highly personalized advertising based on a user’s characteristics and the surrounding conversation.

That possibility goes beyond choosing which pre-made advertisement someone sees.

The system could potentially create commercial language specifically for the context of that conversation.

Another researcher featured in the report, Ines Zelch of Friedrich Schiller University Jena, described experimental work examining advertising integrated directly into generated text.

Her concern is straightforward.

People may not recognize commercial influence when it appears inside the answer itself.

That is where my eyebrows go up.

Please Just Tell Me When You Are Selling Me Something

I do not think every advertisement represents some great ethical crisis.

Sometimes I discover things through advertising that I genuinely want.

I run a media platform. I understand the economic reality that content and technology cost money to produce.

The problem begins when commercial influence becomes difficult to identify.

Transparency gives people agency.

If something is sponsored, label it.

If a recommendation came from a commercial placement, make that relationship obvious.

If conversation data influences ad targeting, tell people what information gets used and give them meaningful control over it.

I should not need a law degree, computer science degree, and 45 minutes with a privacy policy to understand the basic transaction.

ChatGPT and Copilot Were Already Testing Different Approaches

The report also documents tests that eyeo conducted on ChatGPT and Microsoft Copilot in August 2026.

According to those tests, the two services presented advertising differently.

ChatGPT displayed advertiser creative in a labeled ad box following an answer.

Copilot used generated copy related to the user’s prompt above sponsored content.

In both cases, the researchers reported that advertisements matched the prompt context rather than changing the recommendation itself.

Sometimes the advertised company matched the brand mentioned in the response.

At other times, an advertiser was a competitor.

There is an important limitation here.

The report explicitly says these observations came from a small number of test sessions using fresh accounts. They demonstrate what researchers observed, not how frequently users encounter those formats across either platform.

That distinction is worth preserving.

Privacy Is Bigger Than Hiding Your Browsing History

I think many of us still have an outdated mental picture of digital privacy.

We imagine browser history.

Cookies.

Maybe location tracking.

Conversational AI introduces another category: information we voluntarily explain because we want better help.

That can include finances, travel plans, work problems, shopping preferences, family circumstances, relationships, or personal goals.

The privacy conversation therefore has to move beyond “Do companies know which website I visited?”

We also need to ask what happens to the context we intentionally provide.

Different platforms have different policies, settings, and business models, so I would never assume one privacy tool solves every part of that problem.

I do use tools such as NordVPN when I want an additional layer of protection for my internet connection while traveling, especially on networks I do not control. A VPN can help protect network traffic and reduce certain kinds of exposure, but it does not prevent an AI service from processing information that you deliberately type into that service.

That distinction is important.

Privacy tools help.

Privacy judgment still belongs to us.

I Am Not Giving Up AI, So I Need Better Rules of Engagement

I use technology because it can make life easier.

I am not interested in pretending the answer to every privacy concern is simply to stop using modern tools.

That is neither realistic nor particularly useful.

Instead, I think we need stronger habits.

Do I really need to share this detail?

Would I care if this information became part of a commercial profile?

Am I interacting with a tool I trust for this particular task?

Have I looked at the privacy controls available to me?

Can I get the help I need without sharing identifying information?

Those questions sound simple, but they create a much healthier relationship with technology than assuming either complete safety or complete danger.

The Advertising Industry Has Always Wanted to Understand Us Better

None of this began with artificial intelligence.

Advertising has always chased relevance.

Newspapers segmented audiences.

Television targeted demographics.

Search engines targeted intent.

Social media targeted interests and behavior.

Conversational AI may give advertisers something even more precise: context expressed directly by the consumer.

That does not automatically make the technology sinister.

But it does make disclosure more important.

The closer advertising gets to sounding like ordinary assistance, the more clearly users need to understand when commerce has entered the conversation.

Creators Should Pay Attention to This Too

This conversation is not only about technology companies.

It applies to people like me who create content and sometimes earn money through affiliate relationships.

If I recommend something in an article and I may earn a commission from that recommendation, you should know that.

At the same time, I do not want every story I write to read like a sales brochure.

That is why I prefer putting affiliate links inside useful context rather than building articles around products simply because a commission exists.

Trust disappears quickly when every conversation feels like somebody is trying to close a sale.

AI companies will have to navigate the same tension at a much larger scale.

The Real Product May Be Trust

Adblock Plus obviously has a commercial interest in this conversation.

It sells and develops tools designed to block advertising and tracking, so its report should be read with that context in mind.

That does not make the findings irrelevant.

It makes methodology important.

The report identifies its research sources, explains the Harris Poll sample, distinguishes its small qualitative platform tests from representative research, and describes how its ad-blocking figures were collected.

That level of disclosure allows readers to judge the claims more intelligently.

I would like the emerging AI advertising economy to operate with the same basic principle.

Tell people what is happening.

Make advertisements recognizable.

Explain how data influences targeting.

Give users meaningful choices.

Then let people decide whether the exchange feels fair.

I Do Not Mind Being Sold To. I Mind Not Knowing It.

That is probably the simplest way I can explain my reaction to all of this.

Advertising is part of modern life.

So is commerce.

So is persuasion.

I do not expect technology companies to build enormously expensive systems without figuring out how to generate revenue.

What I do expect is the ability to tell when the person, platform, article, or assistant talking to me has a commercial interest in the answer.

Because once I understand the relationship, I can evaluate the recommendation accordingly.

AI assistants are becoming remarkably good at sounding conversational.

The challenge now is making sure the business model does not become so conversational that we stop noticing when somebody has joined the chat to sell us something.

You can read the methodology and findings in the full Adblock Plus report on advertising inside AI.