OpenAI has begun introducing ads to ChatGPT and AI avatars now sell their wares from LinkedIn to YouTube. Consumers must grapple with an unprecedented change to how goods and services are bought and sold. Meta has not only opened up chatbots to targeted ads, but also prevented Meta AI users from opting out.

So how do Americans feel about this shift?

As with most issues, analyzing public opinion does not yield a simple “yes” or “no” answer. Advertising is quintessentially American, after all. While some consumers accept them as a necessary evil, others eagerly seek out and share creative ads, or at least don’t mind fast-forwarding through them.

However, AI tools can cross into personal spaces, and even private ones—from high school journaling to planning vacations. Many users experience AI chats as closer to therapy than search, so ads can feel acceptable in transactional moments but aggressive in reflective or emotional ones. Imagine your therapist taking a break from listening to sell you a supplement.

Using data generated from LLM queries, AI responses will mix real information with sponsored content, and the line between “info” and “ad” may be blurred. The scale of LLM advertising will be unprecedented because it fundamentally alters the ad business. Moving from keyword-based, demographic targeting in Google Search to intent-driven targeting based on conversational context, LLMs are already integrating ads directly into natural language answers as a new “discovery.”

If that is confusing or concerning, you’re not alone. Consumers need to remain skeptical of “free” AI that demands unlimited attention. After all, “free” AI isn’t actually free; it is paid with our attention and data.

Consumers can reasonably expect to decide when an AI tool is allowed to sell to them and when it is not. And most would prefer that Silicon Valley lets them opt in before being sold.

At this point, 41% of U.S. adults call AI ads intrusive, while 33% say it depends on the level of targeting personalization, according to the latest Outward Intelligence survey of U.S. adults. Combined, over two-thirds of responses hinge on how ads are delivered, not whether they exist. In other words, consumers are not against LLM ads in theory, but they are very particular about how they pop up.

The underlying tension is the perceived loss of control that users feel when algorithms decide what’s relevant or not. This isn’t normal advertising; it feels like a loss of agency, and customers are justified in feeling this way. When a user searches for medical advice, only for an intimate question to trigger a “sponsored” response, it can be deeply unsettling.

Even more unsettling is unknown of data changing hands. Is it the AI company keeping the data? A national hospital network? The pharmaceutical industry?

Most consumers want clear boundaries between conversation and commerce. That difference needs to be respected. An AI that treats every conversation as a sales opportunity is not just annoying; it is untrustworthy. And customers may well shop around for the most trustworthy option.

People want AI tools to be both intelligent (requiring data) and private (limiting data use). Ads based on conversation context are the most relevant to large language models like ChatGPT, but they are also the most invasive. 

Companies that solve this paradox through user control, clear labeling, and restrained frequency will capture the persuadable middle. Those that prioritize short-term ad revenue over credibility and trust will lose users before monetization scales.

The 26% of consumers who find AI ads helpful see value exchange: Relevant suggestions in return for attention. However, they are outnumbered by those who fear manipulation or interruption. Nearly half of consumers believe commercials would undermine AI chatbot credibility.

What these findings mean for AI companies is that context-based targeting—say, an ad suggestion based on recent interest in wedding photography—is essentially a coin flip with a lot of money on the line.

At stake is the LLM’s existential relevance. In an inflationary economy, price sensitivity is very real. Only 8% of consumers would pay monthly for ad-free AI, reflecting cost-consciousness but also uncertainty about whether these tools are essential enough to warrant recurring costs. The 40% choosing “free-with-ads” aren’t necessarily ad-tolerant; they may just be unwilling to commit financially to a technology still proving its worth.

Clearly labeled advertising shows honesty about commercial intent. Disclosure preserves trust, whereas hidden or disguised ads will trigger a backlash. Even those who trust monetization more than others will respond viscerally to being deceived.

Outward Intelligence research shows that most Americans are overwhelmed by the age of information—from news consumption to technology on-demand. AI users fear being bombarded, not just tracked. With each individual ad, the risk of reduced AI use rises.

So what’s the upshot? One-size-fits-all advertising will usually fail, and segmentation is critical. Offering opt-out controls or ad-free tiers for the most privacy-conscious users can prevent churn while monetizing the willing majority.

For AI companies, strategy built on public opinion will win out. And it will create winners out of consumers, millions of whom still need to be sold on the idea of AI advertising in the first place. When it comes to AI, there is no free lunch.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.