In the latest episode of the OpenAI Podcast, Asad Awan, Head of Advertising and Business Operations at OpenAI, explained for the first time why ChatGPT incorporates advertising, in what form the ads will be presented, which users will see them, and how OpenAI uses a set of clear principles and mechanisms to adhere to the bottom line of "no monitoring, no interference with answers, and maintaining trust," thus avoiding a repeat of the privacy controversies and trust erosion that are familiar to outsiders.
Ads will only appear on the free and Go plans; they will not appear on the Plus, Pro, and Enterprise plans.
Awan stated that ChatGPT ads will only appear on free users and Go plan users; Plus, Pro, and Enterprise versions will not be displayed.
The company operates three product lines: enterprise clients, subscriptions, and mass consumer products. Enterprise clients and paying users each have their own business models. For most users, the purpose of advertising is to support a viable approach of "high usage, free use," rather than to restrict usage quickly.
Awan pointed out that OpenAI's mission is to make "the best AI" available to more people. Without advertising, the free plan would have to limit usage or only provide weaker models. Introducing advertising, on the other hand, would allow free users to access a more complete and higher-specification service.
With trust as the core principle, users should not feel spied on or monitored.
In response to external concerns that personalized advertising might make users feel monitored, Awan stated bluntly that no matter how good the results are, if it makes people feel uneasy about "being eavesdropped on or monitored," it will not be accepted.
Therefore, OpenAI has clearly defined priorities internally, namely:
"User trust takes precedence over user value, user value takes precedence over advertiser value, and revenue comes last."
Awan pointed out that even if short-term revenue might be higher, the design would not be adopted if it damaged user trust.
OpenAI completely separates models from ads, excluding sensitive conversations from ad delivery and matching.
Awan emphasized that the model's training and responses will not change due to advertisements, and the model itself will not know whether an advertisement is displayed on the screen; visually, the answer area and the advertisement area will be clearly separated. If a user wants to inquire about the content of an advertisement, they need to provide the advertisement information to the model themselves; otherwise, the model will not know that the advertisement exists.
Furthermore, conversations involving sensitive topics such as health, politics, and violence will not display advertisements or be used for ad matching. The relevant definitions and judgments are based on a high-standard classification mechanism from our internal policy team and models, and will be continuously adjusted and tested.
OpenAI will not disclose user conversations to advertisers; users can manage their privacy themselves.
Awan pointed out that advertisers cannot see the content of user conversations. Ad matching is done by OpenAI's internal system, with the goal of providing ads that are "helpful" to users, rather than pursuing a large amount of exposure; if a suitable ad cannot be found, it is better not to display it.
On the user control side, OpenAI provides users with options such as viewing which data is used for ad personalization, choosing whether to use past conversations, clearing history, and turning off personalization; if they don't want to see ads at all, they can upgrade to Plus or Pro. Awan also admitted that this kind of highly controllable and clearable design is not common in the current advertising industry, but is considered a necessary condition for building trust.
OpenAI's future direction is towards agency-based advertising, while emphasizing transparency and user control.
Speaking of long-term vision, Awan described that future advertising may be closer to an "agency" model, helping users compare prices, find deals and suitable products; for small and medium-sized businesses, it may be possible to set advertising goals directly through dialogue, lowering the operational threshold and eliminating the need to rely on complex advertising expertise.
In response to the opposition to "no ads," OpenAI believes that the distrust of online advertising has historical roots, and the company needs to respond with clear principles, transparent mechanisms, and user control. At the same time, it should retain the option of "paying for no ads" so that users with different values can choose the appropriate usage mode.
This article, "OpenAI Discusses ChatGPT's Advertising Mechanism: Paid Users Unaffected, No Monitoring of Conversation Content," first appeared on ABMedia, a ABMedia .
