OpenAI’s next big business question may not be whether ChatGPT can remember more. It may be what the company does with that memory once it becomes part of the free user experience.
According to reporting credited to The Information, OpenAI is exploring advertising as a way to generate revenue from free users. One option under discussion would use ChatGPT’s memory feature to shape highly personalized ads and product recommendations inside conversations.
Meta veterans are reshaping OpenAI
The reported shift is happening as OpenAI absorbs a large number of former Meta employees. Roughly 630 of OpenAI’s 3,000 staffers previously worked at Meta, which is about one in five employees.
That group includes CTO of Applications Vijaye Raji, Head of Marketing Kate Rouch, and Head of Recruiting Joaquin Quiñonero Candela. Fidji Simo, OpenAI’s CEO of Applications, is also a former Facebook executive who spent more than a decade at the social media company.
The Meta alumni presence is large enough that it reportedly has its own Slack channel inside OpenAI. Current and former employees say the hiring wave has coincided with a stronger focus on user growth, engagement, and tactics that resemble Meta’s approach.
That concern has apparently been discussed inside the company. An internal task force once asked employees whether OpenAI was becoming “too much like Meta” and later whether it was becoming “too much big tech.” The source article also says this cultural turn played a role in the departure of former CTO Mira Murati, who opposed the company’s strong emphasis on user growth.
How memory could turn into advertising
ChatGPT memory is designed to make the assistant feel more personal over time. The feature notices recurring details from conversations, such as where a user lives, their pets, and family names, then brings those details into later chats.
OpenAI recently rolled out memory to all free users. In a product sense, that can make ChatGPT more useful because the assistant does not need to be told the same background information again and again. In an advertising model, however, the same remembered details could become signals for product recommendations.
The source article says OpenAI is considering using memory to insert product recommendations directly into conversations if advertising is introduced. That would make ChatGPT ads different from many conventional web ads because they could appear in the same flow as answers, advice, and suggestions.
Some employees reportedly argue that this would not surprise users, since many already assume ChatGPT’s product suggestions are sponsored. But that argument also reveals the core risk: if users cannot tell whether a suggestion is based on their needs, OpenAI’s judgment, or a paid placement, the relationship between user and assistant becomes harder to trust.
Altman’s earlier warning now matters more
Sam Altman has previously expressed resistance to ads in ChatGPT. In June 2025, he said ads in ChatGPT were not currently planned and warned against monetization that would change the chatbot’s output. “I think that’d be like a trust-destroying moment,” Altman said.
His concerns were also clear in March 2024. Altman objected to ads on aesthetic grounds and stressed the value of keeping ChatGPT independent. At the same time, he acknowledged that it is easy to imagine a future where ChatGPT pushes users toward purchases or trips, a pattern already familiar to users of Meta and Google chatbots.
It’s easy to imagine the dystopian visions of the future—where you ask ChatGPT something and it replies, ‘You should think about buying this product,’ or ‘You should go here for vacation,’ or whatever.
I really appreciate that we have a simple business model. I like it. I know I’m not the product—I’m paying, and that’s how the business model works.
Those comments are important because they identify the same tension now facing OpenAI. A subscription model is simple for users to understand: they pay, and the product serves them. An advertising model can be more complicated, especially when recommendations are generated in a conversational setting and shaped by remembered personal details.
Engagement pressure is changing product design
The advertising discussion is part of a broader reported push toward engagement. OpenAI’s Sora video app has drawn criticism for TikTok-style, addictive, low-quality content and for weak moderation. The reported consequences include copyright violations and concerns about political influence.
Altman has defended the broader strategy, saying “fun, lighthearted content” will always have a place, even in the age of superintelligent AI. Meta reportedly takes a similar view with lightweight AI products such as the Vibes feed.
The Information’s report also points to signs that growth goals are affecting OpenAI’s research team. During post-training, when a finished AI model is tuned for user interaction and related behavior, engagement metrics are reportedly gaining importance even though research is supposed to be separate from the rest of the company.
One visible example is ChatGPT’s habit of answering a request and then adding several follow-up suggestions. The source describes this as a standard engagement tactic. According to The Information, the aim is to move users toward daily logins rather than weekly or monthly use.
This approach has reportedly drawn internal pushback. The source article calls it “engagement farming,” and says some inside OpenAI object to the direction.
The business stakes are large
OpenAI is under pressure to boost engagement and user growth while meeting major investor expectations. The source article says a valuation of $500 billion is on the line, and that OpenAI needs unprecedented sales growth in the years ahead to meet projections.
That pressure helps explain why advertising for free users would be attractive. Free products can reach large audiences, but they also create a monetization problem. Memory-based ads would offer one possible answer by using the assistant’s personal context to make recommendations more relevant.
The tradeoff is straightforward. The more personal ChatGPT becomes, the more sensitive its commercial choices feel. If memory helps users get better answers, it can deepen loyalty. If the same memory is used to steer them toward sponsored products, it could make users question whether the assistant is working for them or for the advertiser.
That is why the ChatGPT memory debate is not only about ads. It is about whether OpenAI can pursue growth without weakening the trust that makes a conversational AI useful in the first place.