The world’s first billboard

Picture this. The year is 1835 and you are walking through New York City. There’s a boy on the street corner selling newspapers for a penny and the stories in them detail the expansion of the city and the industrial and religious tensions that go with it.

Not only are the newspapers themselves a novel innovation, but something else catches your eye. From a distance, you can see a massive sign in the street. It is just like the ones you find outside shops but there is no store associated with it. What is more, you can read it from hundreds of feet away and it is promoting a circus.

If you happened to be a US user of ChatGPT on January 9, 2026, you might have seen something similar; the first ads to run on the platform, which could change the marketing and AI landscapes forever.

Clearly, the decision to offer advertising is OpenAI’s way of attempting to generate some sustainable revenue while not destroying the trust it has built up across its user base. But the implications are already echoing across the marketing plans of engineering, technology and scientific companies around the world and there’s a good chance you don’t yet have a line in your budget for it.

OpenAI has moved from advertising theory to practice, and its tests are currently focused on free and low-cost tiers, with ads positioned below responses and clearly labelled to distinguish them from the AI assistant’s output.

At present, paid subscription plans, which means Plus, Pro, Business and Enterprise, are ad-free; but that’s very unlikely to remain the case. After all, whoever heard of an advertising entity permanently denying its customer’s access to its wealthiest user bases?

This approach lets OpenAI test out some advertising formats and find out what its users will tolerate without risking the revenue stream it has built up among subscribers. In other words, it allows it to get good at what it’s doing, before it starts to annoy the people putting their money where its mouth is.

The biggest noise around ChatGPT’s ad launch has been whether it will create a compromised service for its users, which they might start to trust less. OpenAI has sought to address this by emphasising that advertising will remain separate from ChatGPT’s core outputs.

Right now, ChatGPT’s responses are generated independently and are not influenced by paid placement, or at least this is what OpenAI would claim, and there seems little reason to argue. That said, it’s entirely possible and entirely legitimate to influence those answers using Generative Engine Optimisation (GEO or AI Optimisation (AIO) — feel free to choose your own acronym.

Furthermore, the company has been very clear that the chats you have with it are not shared with advertisers and are not sold as third party data. For me, this is wonderful news, because you have no idea how many inane questions I ask.

The ads that are running in the trial are designed to be clearly identifiable as promotional. The intention is that users can distinguish between the GPT’s responses and advertiser’s work, but my own experience of being inside the marketing industry looking outwards, tells me that many people’s radars aren’t quite as attuned to ads as we would like to think.

Users are also expected to retain control over their experience, including the ability to dismiss ads, understand why they are seeing them, adjust personalisation settings, or opt out entirely by subscribing to an ad-free tier. You will be familiar with this functionality; it’s all the stuff you currently don’t do on Instagram, TikTok, LinkedIn and Facebook; if that’s still a thing.

To support this shift, OpenAI has been investing in internal capability, including teams dedicated to ad integrity and safety. These groups are responsible for ensuring advertisements meet strict standards for relevance, accuracy and appropriateness, and that they do not undermine trust in the platform.

All this work, and heavy investment, suggests that advertising is not just an experiment, but a strategically important part of OpenAI’s long term business plan. From an outsider’s point of view, they need to do something soon if they are going to keep investors interested and avoid the much-vaunted bursting of the AI bubble.

The introduction of ads has also sharpened competitive positioning within the AI market. Rival platforms have been quick to contrast their own ad-free approaches, framing them as more privacy first, user centric alternatives.

In fact, Anthropic recently ran a Super Bowl ad to tell people that its own AI, Claude, was ad free. Hang on, they ran an ad to tell us they were ad free…

Despite my cynicism, these ads are hilarious. Try this one, in which a man asks how to connect better with his mum and the chatbot, personified by a motherly older woman, asks if he would like to start dating cougars. (That, by the way, is the first article I’ve written which hyperlinks the phrase ‘start dating cougars’.)

If cougars aren’t your thing, you might enjoy this work of art, in which a young entrepreneur is offered a loan ‘because girl bosses need SHE-E-O money quick”.

Sam Altman, the OpenAI CEO, wasn’t amused though. “We’re not stupid,” said Altman in an interview with the TBPN podcast. “We respect our users, we understand that if we did something like what those ads depict, people will rightfully stop using our product.”

This has exposed a growing, and potentially existential, divide in the sector. Should chatbots, such as ChatGPT, look like traditional ad-funded platforms, or would subscriptions and enterprise funding provide a more honest and cleaner approach?

If, like me, you remember other search engines claiming they wouldn’t run ads when Google unleashed its first sponsored positions back in 2000, you probably feel the same about the long-term future of an ad-free AI assistant. Once a price point is established, users are very unlikely to spend a great deal more to get a truly ad free experience.

There are plenty of people out there, in the analyst-verse, that see OpenAI’s move as pragmatic. Running large-scale generative AI systems is extremely costly, and advertising offers an easily understood and potentially scalable revenue stream.

However, sceptics argue that advertising alone is unlikely to resolve long-term cost pressures, and that OpenAI will need a diversified mix of subscriptions, enterprise services and platform partnerships to remain financially resilient.

“Profit is the metric that ultimately counts,” wrote marketer and academic Mark Ritson in AdWeek. “Deutsche Bank estimates the company will accumulate $143 billion in negative cumulative free cash flow between 2024 and 2029. That’s fancy Wall Street talk for a giant hole.”

Of course, I also remember people saying Google would never work unless it launched a proper homepage with chat forums and email on it like America Online had, and we all know where that ended.

For advertisers in the scientific, technology or engineering worlds, ChatGPT represents a fundamentally new environment. Ads delivered in a conversation with an LLM could reach your audience at moments of high intent, when they are actively researching, comparing or problem solving.

My own theory is that, while AI search has dragged traffic away from websites around the world, including, I would hazard a guess, your own, the buying is still done on commercial websites.

This creates a wonderful opportunity to do what I have argued, for well over a decade, that B2B brands should advertise their expensively produced and maintained knowledge bases. Afterall, if someone is in the research phase, then your white paper, eBook, product comparison tool or blog might be something they would value.

Overall, getting in on the act early might offer B2B organisations the chance to experiment with lower competition and get ahead of the pack.

However, at the time of writing OpenAI.com states, “There are currently no ads in ChatGPT,” before going to say, “At this stage, we’re focused on learning from early testing in the coming weeks. We’ll have more to share as we learn from early use and feedback.”

In a setting which, in theory, is built on trust and usefulness, promotional messages that feel intrusive, like Claude’s cougar, or ones that are poorly aligned with what the audience wants, like the SHE-E-O loan offer, are going to tank. In contrast, companies that use ChatGPT to share genuine and valuable information and offer solutions stand the best chance of long-term success.

Having said that, there is always the argument that being the only person in the library shouting can be a pretty effective way of getting attention. I did used to get thrown out of a lot of libraries though…

As testing continues, the coming months will reveal whether ChatGPT can successfully balance monetisation with credibility, and whether advertising in AI assistants can become both effective and acceptable at scale.

My feeling is that, if you have seen an ad on ChatGPT, you should screen grab and date stamp it now to show your grandchildren; you’ve just witnessed the birth of the billboard.

We get that it’s hard to keep up with the latest thinking in PR and marketing. Subscribe to our twice weekly blog, direct to your inbox, for the answers to all the burning questions your board ask – from using AI in marketing to how to spend your budget against a backdrop of constantly multiplying channels.