Ads in AI: the death of impartiality?

What happens when the tool you trust for neutral answers starts recommending products?

AI tools like ChatGPT are, for many people, becoming the main interface for finding information, making decisions, and getting work done. As adoption grows, so do the costs of running advanced models. OpenAI, as a high-profile example, burns through an estimated $17 billion per year. Now advertising is entering the picture, raising a fundamental question: can AI remain trusted if it becomes ad-supported?

Some companies believe advertising is necessary for scale. Others argue it risks damaging the credibility that makes these tools worth using in the first place. Here is the case on both sides.

Why advertising in AI makes sense

Running large language models is expensive. Training, infrastructure, and ongoing usage cost hundreds of millions of dollars each year. Subscription revenue alone may not cover long-term demand, particularly for companies that want to offer free access to as many people as possible.

OpenAI has begun testing adverts in ChatGPT for free users. These placements are clearly labelled and appear within the chat interface, with the stated aim of supporting operating costs without placing the full burden on paying subscribers. Early advertiser partners include Target, Adobe, and Ford, all of which see value in reaching users while they are actively asking questions or completing tasks.

Image source: OpenAI

There is also an economic accessibility argument. If ad revenue supports a free or lower-cost tier, more people can use AI tools without paying subscription fees. Students, job seekers, and small businesses stand to benefit most from that arrangement.

AI also opens up genuinely new advertising formats. Instead of static banners, conversational ads could respond directly to user intent. Someone asking about travel plans might see relevant hotel offers. A user designing a logo might be introduced to creative software. In theory, this increases relevance and reduces wasted impressions for advertisers and users alike.

Conversational advertising could be the most contextually relevant ad format ever built. The question is whether users will accept it inside a tool they trust for honest answers.

The concerns about adverts in AI

Trust and neutrality are central to the appeal of AI assistants. That is precisely what makes advertising such a sensitive issue here.

Perplexity removed advertising after testing, citing concerns that ads could undermine trust in factual and unbiased responses. Anthropic has publicly pledged that Claude will remain ad-free and has promoted this position as a differentiator, making the contrast with competitors explicit in its Super Bowl advertising.

Critics argue that advertising may shift priorities toward commercial interests rather than accuracy and quality. If a platform earns revenue from sponsors, users will reasonably question whether recommendations are fully impartial. Academic research has suggested that even the perception of advertising influence within AI conversations can damage trust, regardless of whether the ads are clearly labelled.

Privacy is another significant concern. Users may worry that their queries are being analysed to personalise adverts, particularly when asking sensitive questions about health, finances, or personal circumstances. The backlash against targeted advertising on social networks and search engines is a warning the industry would be unwise to ignore.

Some users, myself included, believe conversational tools should remain ad-free altogether. The frustration with intrusive advertising that has built up over two decades of social media and search is real, and AI assistants have so far avoided it. That is worth something.

We have already seen what happens when advertising becomes the primary business model for tools people rely on daily. The incentives shift, and over time, so does the product.

Where the industry stands now

Different companies are taking meaningfully different approaches. OpenAI is actively testing adverts in ChatGPT to support free access. Anthropic’s Claude and Perplexity are both positioning ad-free experiences as a core part of their value proposition. Google’s Gemini has not announced advertising plans at this stage, though it sits within a company whose entire business is built on ad revenue, which is a situation worth watching.

The industry is still in early experimentation. It is likely that both ad-supported and ad-free models will coexist for the foreseeable future.

What this means

Advertising can make AI more economically accessible and create new revenue streams that keep capable models free to use. It may also enable conversational ad formats that feel more relevant than anything that came before.

The counterargument is equally strong. Advertising can erode trust, raise privacy concerns, and alter perceptions of neutrality in ways that are difficult to reverse. Because AI assistants operate as reasoning partners rather than simple search engines, the stakes are considerably higher than they were for web search.

The outcome may depend less on whether ads appear at all, and more on how transparently and carefully they are implemented. But transparency and commercial incentive do not always travel well together. The industry will need to watch that carefully.

Frequently asked questions

• Does ChatGPT show adverts?
  OpenAI has begun testing adverts in ChatGPT for free-tier users. The placements are labelled and appear within the chat interface. This is currently being trialled rather than rolled out universally.
• Does Claude show adverts?
  No. Anthropic has publicly committed to keeping Claude ad-free and has positioned this as a deliberate strategic choice rather than a temporary one.
• Can advertising affect the neutrality of AI responses?
  Research suggests that even the perception of advertising involvement can reduce trust in AI responses, regardless of whether individual answers are actually influenced. Transparency about ad placement helps, but does not fully resolve the concern.
• Why are AI companies considering advertising?
  The cost of running large language models is substantial. Advertising offers a revenue model that can support free-tier access without relying entirely on subscriptions, which makes it commercially attractive despite the trust implications.
• What is the alternative to ad-supported AI?
  Subscription models, enterprise licensing, and API pricing are the main alternatives. Some providers also explore revenue sharing with developers who build on their platforms. Each model involves trade-offs between accessibility, cost, and the incentives it creates.