AI assistant ads: What consumers think about advertising in LLMs

Ads are coming to AI assistants. The only real question now is whether users will accept them. 

To understand where consumers stand, we surveyed 1,000 LLM users in the US to uncover what today's early adopters are telling advertisers where they will (and won’t) accept ads in AI assistants. 

Read on for some of our key findings.

While 82% of LLM users shared that they would find AI assistant ads at least as trustworthy as Google Search ads, that starts to vary when broken down by age group.

Among 18–34 year-olds, 42% shared more trust in AI ads than Google Search ads. For those ages 35–54 it dips to 31% and among 55–75 year-olds, it drops to just 17% — with two-thirds of older users saying trust would be “about the same.”

This variation in trust levels is likely due to younger users living more natively in conversational interfaces like these. They treat AI as a more authoritative source — closer to a recommendation than a placement.

When it comes to the level of annoyance these ads would bring to users, there isn’t a vast difference compared to what’s already out there. 

42% of users place AI ads at the same annoyance level as social media ads, with 25% saying they would be less annoying and 33% saying more.

Social media is the volume channel most media buyers know best. The data says AI assistant ads land in the same emotional zone, so it makes sense users would feel about the same.

Naturally, the introduction of ads is a big deterrent for some. But people are willing to pay to avoid seeing them.

In fact, 40% of users shared they would be willing to do so — with 22% saying they would pay for an ad-free version and 18% sharing they would switch to a different AI assistant if ads were introduced. 

Among paying users, 35% would pay to escape — more than double the 17% among free users.

This makes sense as free users are typically more tolerant, while paying users are the audience most worth winning (and most ready to leave).

It’s context, not volume, that is also drawing the line for what's acceptable among users. 

There are five distinct contexts where users find ads most unacceptable:

These five contexts combined account for 70%+ of consumer concern, which means that context exclusions should be treated as non-negotiable in any AI assistant deal.

Unsurprisingly, paying users will likely be the ones drawing the hardest lines on what’s acceptable here.

When it comes to confidential work context, 22% of paying users say ads are unacceptable, compared to 14% of free users. Regarding medical context, 18% of paying users say ads are unacceptable vs 12% of free users. And in a finance and legal context, it's 13% versus 10%.

Paying users aren't generally anti-advertising, but they're clearly specifically anti-advertising in high-stakes contexts.

If ads are introduced, 27% shared they would turn off memory and personalization entirely. While 25% would keep memory but opt out of ad personalization. And 15% shared they have already had these features turned off.

This isn't hostility — it's a rational privacy response. And again, paying users are the most surgical here, with 31% saying they would opt out of ad personalization specifically.

All of this said, users still aren’t ready to take a firm stance. 

From our data, there are six different views fragmented across users: 

• 22% say ads are fine if the tool stays free. 

• 18% say ads make AI feel less trustworthy. 

• 18% say ads in AI feel the same as ads anywhere online. 

• 18% say it is more invasive than other digital ads. 

• 15% say no impact. 

• 11% would rather pay than see ads.

This means that the narrative is still being written, so it's up to advertisers how to move forward. 

Our recommendations? 

1. Get in early. The market is forgiving — but the window is closing. The most engaged users will progressively self-select onto ad-free alternatives. Whatever you ship in 2026 will shape what consumers expect from your category in 2028.

2. Make context exclusions non-negotiable. The five concern contexts — confidential work, mental health, medical, child usage, financial — are bright lines. Push platforms for transparent, auditable context detection. Treat the absence of clear exclusion tooling as a deal-breaker.

3. Build for the noisier signal. Half your target audience will withdraw their personalization data once ads appear. Invest now in contextual targeting — semantic, intent-based, query-derived. The behavioral floor in AI assistants will be lower than anything you're used to.

Consumers aren’t drawing a hard line against ads yet — but they are making it very clear where the boundaries are: high-stakes conversations and overly personalized targeting all trigger stronger resistance, especially among paying users.

The leaders in AI advertising will likely be the ones that understand the nuance of conversational environments and treat them differently from search and social feeds. Because once ads enter AI assistants, brands won’t just be competing for attention like in other channels. They’ll be competing for trust.