It's Not the AI, It's the Suspicion of Influence: What the Ipsos Numbers on Advertising Show

The Ipsos AI Monitor 2026 delivers a picture that looks contradictory at first glance: more than half of Germans already use AI tools in everyday life, while general trust in the technology drops sharply. Three in four respondents demand clear labeling of AI applications. But the actually interesting figure is more precise: more than half of consumers would trust generative AI results less the moment advertising influences them. For general brand content, by contrast, approval and rejection sit almost evenly split.

The number that's surprising

That gap deserves more attention than the headline "AI advertising loses trust." It isn't advertising itself that bothers people – for classic brand content, sentiment is nearly balanced. It's the specific idea that an answer looking like neutral information might have been altered by commercial interest. That's a different, more specific concern than general tech skepticism: it isn't directed at AI, but at the possibility that a system presumed neutral gives up that neutrality for payment without anyone noticing.

Why this confirms my own thesis

That's exactly the concern I traced back to ChatGPT advertising's exclusion list four days ago: that OpenAI itself exempts certain conversation topics from ads was already a sign back then that the trust context in a conversation differs from a search query. The Ipsos numbers now provide quantitative confirmation of that assessment, independent of any single product: it isn't the individual platform that has the problem, it's the format itself – an answer that looks like information but might be advertising.

What the study itself leaves out

The same sober scrutiny that this finding deserves applies to the study as well: the underlying report states neither the exact sample size nor the precise survey period. That doesn't make the numbers wrong, but it means reading them as a directional signal, not a precise measurement – the same caution I already flagged regarding unsubstantiated efficiency claims applies just as much when a number happens to support your own argument. A study that demands caution toward unsubstantiated claims should disclose what it's based on itself.

What this means for campaign work

In practice, this means: whether an AI platform technically separates ads from answers is no longer a detail of system architecture, but the central trust question every ad in this environment hangs on. A brand advertising inside an AI assistant isn't just selling a message – it's implicitly vouching for the claim that the message hasn't altered the actual answer. A campaign that ignores this distinction and simply transplants existing search-ad logic works directly against the exact concern more than half of respondents name.

Conclusion

The Ipsos numbers don't show blanket AI distrust – they show a precise concern: that an answer looking like neutrality might be for sale. That confirms what ChatGPT advertising's exclusion list already hinted at, now with real numbers instead of a single observation – even though the study itself should have disclosed exactly what those numbers are based on. Advertising in this environment means competing not just for attention, but for proof that nothing was changed.