American artificial intelligence (AI) company Anthropic this month attracted applause – and a surge in users – for clever advertisements poking fun at its competition.
In the commercials, an AI assistant awkwardly breaks away mid-conversation to push products such as shoe insoles and dating services. “Ads are coming to AI”, the Super Bowl-tied spots warned, but not to Anthtropic’s own chatbot Claude.
The campaign quickly generated buzz because it played to peoples’ worries that inviting advertising into AI platforms which many of us now rely on – and confide in – risks blurring the line between helpful advice and paid influence.
But that anxiety, while understandable, overlooks how advertising already works across much of the digital world.
In many ways, ads based on our interactions with AI aren’t such a big leap from the kinds of targeted advertising that already dominate search engines, social media feeds and e-commerce platforms.
And if transparent and well-designed, the shift could help people complete tasks faster and keep these tools widely accessible.
AI’s access and equity headache
This month, OpenAI’s ChatGPT began testing adverts with users in the United States. The company assures us any ads will be clearly labelled, kept separate from answers and accompanied by privacy protections and user controls.
The stakes are high: ChatGPT now boasts 800 million weekly users and ranks as the internet’s fifth most visited website. It has operated largely ad-free since its launch three years ago and only about 5% of users pay a subscription.
With room to grow, OpenAI has strong incentives to find a sustainable model that protects trust without undermining what made the service so popular.
If indeed transparent and optional, its advertising could help solve a basic funding problem. In practice, a small paying group cannot carry the full burden forever.
A light, clearly labelled ad model is one way the wider user base could contribute indirectly – much as they already do via television, YouTube, search engines and many news websites.
That matters for access. Around one in six people worldwide already use generative AI, but adoption is uneven and a digital divide is widening between richer and poorer countries.
If wealthier nations move faster, sustainable business models can help spread access by keeping costs down for students, job seekers and small organisations in emerging economies.
The convenience of ‘contextual’ advertising
For everyday ChatGPT users, the main upside of ads is that they can reflect what is needed in the moment, rather than what a tracker infers from past browsing.
Traditional digital ads use cookies and cross-site tracking to guess people’s interests over time. Contextual advertising, by contrast, targets what is happening on the page or in the moment and is often seen as a more privacy-friendly alternative.
OpenAI says ads will be matched to the conversation and may use past chats and ad interactions. Users will be able to dismiss ads, see why they were shown one and delete ad data.
If those controls work as promised, relevance would come from the question being asked, not from tracking across other websites. Imagine asking: “I’m hosting friends. What are two easy Mexican dishes, and what ingredients do I need?”.
ChatGPT could give the recipe guidance first, then show a clearly labelled ad option, such as a local supermarket delivery link for the exact ingredients, or a sponsored meal kit that fits the budget and dietary needs. Instead of jumping between tabs, the user moves straight from decision to action.
For consumers, that is convenience. For advertisers, it is also efficiency, because the ad appears at the moment of genuine intent rather than being sprayed across the internet.
Another benefit is smoother communication. Conversational ads have the potential to function more like a shop assistant than a static banner. Instead of clicking away, opening tabs and filling in forms, follow-up questions can be asked in the same chat and personalised details returned quickly.
OpenAI suggests this could include sponsored listings that users can interact with in the chat. For instance, while planning a trip, a sponsored accommodation option might appear, allowing questions about availability, cancellation, location and total cost for specific dates and group size to be handled in one place.
Done well, this could reduce frustration and curb misleading advertising, because people can challenge vague claims and ask for specifics before spending money.
Trust, transparency and limits
None of this removes the risks. Advertisements should not be allowed to change what a trusted AI tool such as ChatGPT recommends. And because ads are currently being tested with only a small group of users, the full extent of those risks cannot yet be observed or properly assessed.
That is why transparency and separation are not cosmetic. They are safeguards.
For now, it may be tempting to treat “ad-free” as the only ethical position, as Anthropic’s new campaign implies. But the world is still early in this shift. These systems should be judged by what happens in practice – especially on transparency, user control and real protections against manipulation.
If those guardrails hold, it is worth considering the upside too: ads in AI tools could support access, reduce friction and help more people benefit from this powerful technology.
