OpenAI is eyeing ads – but getting it right will be a tricky balancing act

It’s a possibility that experts say could help drive revenue and aid the company’s shift to a for-profit model – but, if deployed poorly, could damage user trust.

The company’s chief financial officer, Sarah Friar, told the Financial Times on Monday that OpenAI is assessing the potential for an ads-based model and that the company would be “thoughtful about when and where we implement [ads].”

Friar, who previously held posts at Salesforce and Square, noted that she and the company’s chief product officer, Kevin Weil, an ex-Instagram executive, bring significant advertising expertise to the table.

The news comes at a critical juncture for the Microsoft-backed company, which in October secured $6.6bn in funding from Microsoft, Nvidia, Thrive Capital, Fidelity, Softbank and others. Now valued at around $157bn, OpenAI is in the midst of a drastic transformation, considering becoming a public benefit corporation that would no longer be under the leadership of a nonprofit board.

And as OpenAI pursues a profit-focused model, it’s facing the challenge of revenue generation. While the company’s monthly revenue reached $300m in August – a lift of 1700% since the start of 2023, according to the New York Times – it’s still burning through cash, projecting a loss of $5bn this year to operating costs and other expenses. More advanced models, such as the premier version of ChatGPT – GPT-4o, released in May – require increased compute power, incurring ever higher costs for the company.

Selling ads within ChatGPT and other OpenAI programs, such as Dall-E and the OpenAI o1 models, could help advance the company’s goal of becoming profitable.

“It’s long been known that OpenAI is burning cash at a crazy rate in order to keep up their operations,” says Christopher Penn, co-founder and chief data scientist at TrustInsights.ai, an analytics and AI firm for marketers. “When you think about it, ChatGPT as a tool is absurdly [low] priced for what it delivers. As they’ve debuted new models, such as the [flagship] o1 model, it’s clear from the cost of tokens in that model that it is a crazy expensive model for them to run. The hardware requirements and the processing power point towards that model being the future of the company, but at a substantially increased price. Advertising allows someone else to pay the bill other than the users.”

OpenAI wouldn’t be the first generative AI leader to pursue an ads-based model. Microsoft’s AI chatbot Copilot (previously called Bing Chat) and AI search engine Perplexity, for instance, have both introduced ads that appear in generated responses to user queries. Meanwhile, Google announced in October that it is debuting ads in its AI Overviews, which provide AI-generated summaries at the top of search results.

While OpenAI is exploring the possibility, Friar caveated in a statement shared with The Drum that the company has no “active plans to pursue advertising.”

A boon to business?

Experts believe that the addition of ad products could be advantageous to OpenAI as it battles rising operational costs.

“It’s hardly surprising that OpenAI would go down the advertising route. Since they announced their for-profit overturn, there’s no question that introducing ads into their products is one of the most straightforward ways to monetize any free software-as-a-service product,” says Eli Goodman, CEO and co-founder at Datos, a Semrush-owned clickstream data provider.

He points to Meta-owned Facebook and Instagram as prime examples of the success of ad-based models. Both platforms introduced ads a few years after their respective launches when they had already developed dedicated user bases. Since OpenAI, too, has a highly engaged user base – now tallying around 200 million active weekly users and more than a million paid business users – the company isn’t likely to repel users with the addition of ads, Goodman suggests.

Others agree with the assessment that ads will prove a promising route for the developer. Dustin Engel, the CEO and founder of marketing consultancy Elegant Disruption, for example, says it would be “hard to imagine a scenario where ads aren’t a critical part of OpenAI’s business model,” especially considering that it needs to compete with other platforms that have already embraced ad-based models.

Engel and others argue that OpenAI’s access to vast swaths of user intent data could potentially offer more precise targeting for advertisers and more relevant recommendations for users. If a language model such as ChatGPT can effectively identify where a user is at in their customer journey – and their level of intent – it will likely be able to surface ads at an opportune moment.

But the success of this approach, some experts warn, hinges on the platform’s willingness to establish responsible, privacy-conscious policies for data sharing. OpenAI could follow the lead of Perplexity, for instance, which includes ads in responses to user searches but promises to never share users’ personal information with advertisers.

Will transparency butt heads with user experience?

Despite the promise of an ad-supported model, a number of hurdles remain.

For one, OpenAI will need to be mindful to disclose advertisements explicitly – and not disguise them as organic responses to prompts.

“Consumers generally don’t have a problem with ads as long as the ads are not deceptive and they’re clearly marked,” says TrustInsights.ai’s Penn. “If OpenAI, in the ChatGPT interface, has a response and then there’s a big, bright yellow ‘here’s an ad’ block that clearly denotes where the generated answer and the advertising-placed answer are, that would be [good]. If, on the other hand, the user can’t tell the difference between a purely generated answer and an advertising-placed answer, that would substantially impact trust because you don’t know whether the answer is actually correct or was paid to be correct. In today’s media environment, where trust is at all-time lows for everything, that could be substantially harmful to their business.”

But other experts argue that ads shouldn’t be highly disruptive – a position that might come into conflict with calls for clear disclosure.

Users don’t want to be thrown off by blaring ads that detract from their experience – on the contrary, ads embedded in generated responses should be additive, argues Craig Elimeliah, chief creative officer at Code and Theory, a digital-focused creative agency. “If OpenAI does include ads, it is important that they don’t feel like ads,” he says. “They have to improve the experience, not [just] ‘dis-interrupt’ it.”

Elimeliah goes so far as to argue that OpenAI should “consider rethinking ads altogether, blending them into what utility users have come to expect.” He urges OpenAI to ditch traditional ad banners and instead embrace “tools or experiences that serve the user.” An example he offers is “curated recommendations that feel anticipatory, personal, helpful and aligned with the prompt.” In his view, “the magic is in making ads feel like they belong and are useful and smart.”

It’s a view shared by Greg Swan, senior partner at marketing agency Finn Partners, who says: “The challenge – and the opportunity – is to ensure that these ads enhance the experience rather than detract from it. If AI ads can be contextually relevant, helpful or even entertaining, they could feel like a natural extension of the platform’s capabilities, much like search ads did for Google, Yahoo and others.”

Like Penn, Swan emphasizes the importance of maintaining user trust through transparency – ensuring that users are never in the dark about whether they’re being served an ad.

Of course, ensuring ads are clearly disclosed while maintaining a seamless user experience is a fine line to toe.

And even if OpenAI is able to develop compelling ad products, winning over ad dollars may prove challenging. OpenAI won’t just have to supply audiences – it will also have to prove out its performance potential.

“Most advertisers are used to paying for eyeballs, or tonnage,” says Penn. “OpenAI would have to convince advertisers that their product performs better not on a traffic perspective, but on a performance perspective – that their language model successfully identifies commercial intent and directs the user when the user is ready to make a purchase. If [OpenAI] can do that, they will [prove to be better for brands than] spending a huge amount on impression-based advertising, say, on Instagram.”

The brand safety conundrum

Another hurdle for OpenAI is ensuring brand safety. Platforms such as ChatGPT can produce unpredictable and controversial content. Further, misinformation and ‘hallucination’ remain unsolved problems in generative AI. In October, Dow Jones and the New York Post filed a lawsuit against Perplexity, alleging that the platform generated fake bits of news reporting and incorrectly attributed the text to the publishers. This phenomenon presents serious concerns for many brands.

“If an ad appears in an inappropriate or poorly contextualized AI output, it could backfire,” says Swan. “OpenAI will need robust brand safety mechanisms and clear content guidelines to mitigate this.”

The sophistication of OpenAI’s brand safety guardrails may depend on the level of integration between the model in question and the ad products. “If the language model can understand intent and make recommendations at the right time based on the conversation, [OpenAI] will do well. If it just is flinging ads on a primitive keyword basis, it’s going to do poorly because that’s no better than what you can get in traditional search,” says Penn.

Ultimately, however, the onus for managing brand safety will remain on the advertisers themselves. Max Kalehoff, chief growth officer at Realeyes, an ad testing firm that uses AI and computer vision to measure attention and engagement, advises that advertisers “closely monitor their presence and reputations, and develop strategies and tools to proactively manage them.”

Of course, once advertisers are enticed, they’ll demand granular performance reporting to justify their spend. OpenAI will also need to grapple with the task of providing focused metrics and proof of return on ad spend to maintain brand investments.

A new era of AI’s interplay with adland

Should OpenAI choose to invest in a hybrid ad- and subscription-based model, it may establish a precedent in the industry.

“The key will be execution,” says Elegant Disruption’s Engel. “If OpenAI missteps and alienates users, others may hesitate to follow suit. Conversely, a well-implemented ad strategy could set the standard and drive broader adoption. Competitors such as Anthropic or MidJourney may feel pressure to adopt similar strategies to remain competitive.”

Realeyes’ Kalehoff is more bullish. “I expect the trend to gain steam. Advertising has long been a large, resilient component of our world economy. The players who dominate control and benefit from advertising come and go, though the advertising economy itself has proven its sustainability over and over again for well over 100 years. AI companies need to make money somehow.”

And as AI capabilities expand, so too do the possibilities – and the risks – for brands.

For example, the novel proliferation of AI agents – who can make decisions and take actions on behalf of users, like, say, making salon appointments or booking dinner reservations – is unleashing new opportunities.

For instance, says Penn, “If you’re on a recipe website and you want to have AI generate a recipe, there are opportunities within that workflow to use [both] AI agents and traditional code to introduce products. If you’re asking for a recipe for a Christmas ham and your language model understands that named entity and you have an advertiser – Honey Baked Ham or whoever – that could be injected in.”

But the possibilities are greater still. As AI agents become increasingly common, brands may be able to market or advertise directly to agents rather than to real users. In response, the agent could organically recommend a product or service to its user. In Engel’s view, “that’s the holy grail – getting your product recommended by an AI that knows its user inside out.”

Meanwhile, the generative AI boom is reshaping the media landscape at large by empowering users to create their own personalized content on demand, from stories to films, in mere seconds. It’s a shift that’s increasingly threatening traditional media’s relevance and challenging advertisers to adapt to a world where broad-reach strategies no longer apply. With users engaging privately with AI platforms to create their own content, ad opportunities tied to public media consumption are shrinking. In this paradigm, Penn predicts, the brands that come out on top will be those that build loyal, direct audiences and work to get their messages seamlessly embedded in personalized AI experiences.

If one thing is clear, it’s that the tides are changing dramatically.

“This shift will push the industry to rethink how advertising and AI coexist, balancing monetization with user trust and creative integrity,” says Engel. “For OpenAI and others, the opportunity is immense – but so are the stakes.”