Adblock Plus and (a little) more
Ads are coming to AI. So far we've blocked 16 million. · 2026-07-09 14:44 by Cornelius
When ads started appearing inside artificial intelligence (AI) assistants, Super Bowl commercials made fun of it and Sam Altman threw a tantrum. Since then, companies like ChatGPT-maker OpenAI have not shared much about how they serve ads to you. AI assistants are becoming the new way people find information online, and we wanted to understand better how ads work inside them. So we started counting how many we’re blocking and developed the AI block-o-meter, a public dashboard on ad blocking and user sentiment toward AI platforms.
What are AI chatbot ads, though? If you haven’t seen one (you’re probably using an ad blocker), they look a little something like this: they’re ads in your chat feed that mostly appear as separate, labeled units, typically at the end of an answer the AI chatbot gives you. Our filters automatically detected and blocked them from day one, but we became curious about them. Were there a lot of ads? Can we quantify how many we’re blocking? So we started counting as of late April and since then, Adblock Plus and AdBlock have blocked 16 million ads on ChatGPT alone. We’re currently implementing the counting system for Microsoft Copilot, the numbers are coming soon.
With the AI block-o-meter, we wanted to create a free and public resource, measuring ad blocking inside AI assistants. It shows you how many ads our products blocked, what trends we see over time, and soon how different AI tools compare. We’ll add more platforms as they start showing ads. This library is, as far as we know, the first of its kind.
User concerns about ads appearing in AI-generated answers
We did this because we know that people aren’t comfortable with ads in AI. 72% of United States (US) adults are concerned that AI chatbots could use their conversation data to target them with ads, while 44% are very or extremely concerned about it. But we wanted to dig a little deeper and understand better how people feel about these ads. We ran a survey asking about user reactions to ads in AI assistants. Most responses were negative and people described the experience as “annoying,” “intrusive,” or “irritating.” Several users specifically said that these ads break their concentration or cause trust issues. One of the respondents actually said that ads simply “should be blocked.” To that person, wherever you are: we’re on it.We’re not counting these ads as a gimmick. We’re counting them because AI chatbots are becoming central to people’s lives and how they get information, but we know almost nothing about how ads inside those tools work, how many there are, and how they are chosen. We’re trying to measure what we can, and we’re doing so in the open.
What we do not know
Of course, it would be interesting to better understand how ads are targeted at individual users or what words in the prompt trigger what types of ads. But this is something that others have to explore, because we never track any uniform resource locator (URL) paths, any of your conversations or prompts, or any other data that could identify you – for obvious reasons. We only count the ads that our public filter list rules block and attribute to one of the known AI domains. We pull the counts from anonymous, aggregated data that includes statistical “noise,” meaning we adjust the data by a tiny random amount. This way, the figures are accurate, while no data point can be traced back to you individually. Finally, we do not believe that ChatGPT and Copilot will remain the only AI tools to feature ads. Some reports suggest that ads are coming to Gemini next. We can assure you that no matter where they show up next, we will block them and share what we learn with you. It’s time to open the black box.