OpenAI ships ChatGPT ads in six markets with a two-tier consent regime

OpenAI has rewritten its privacy policy to cover ad personalization and expanded sponsored content in ChatGPT beyond the United States, according to a report from cryptobriefing.com crediting gizmodo.com ^[1]. That moves ChatGPT out of the pilot category and into the list of ad channels a media planner has to price, with the added complication that its consent rules differ by jurisdiction ^[8][9].

The sequence matters. Testing began in the US on February 9, 2026, aimed at adult users on the Free and Go tiers, with Pro, Business, and Enterprise kept ad-free ^[2][3]. By August 11, 2026, the program covered the UK, Mexico, Brazil, Japan, and South Korea ^[4]. That is roughly six months from first test to a six-market footprint, counting the US ^[5][6]. The policy work tracked the rollout rather than preceding it: US ad-related data handling updates landed around April 30, 2026, and a separate update on June 2, 2026 addressed EU and UK requirements ^[7].

The consent split is the part that changes campaign math. In the US, ads are served by default and users must opt out ^[8]. In the EU and UK, personalized ads require explicit consent before they can be shown ^[9]. Cryptobriefing's read is that the US model leaves most users personalized unless they go into settings and flip a toggle, while the opt-in requirement will likely produce lower personalization rates among European users ^[10][11]. Nothing in the source specifies which regime applies in Mexico, Brazil, Japan, or South Korea ^[12], so anyone building a cross-market plan is working with one confirmed opt-out market, one confirmed opt-in market, and four unspecified ones.

On the data side, the updated policy names chat context, ad interactions, and, for eligible users, stored memories in ChatGPT as inputs that can be used to serve relevant ads ^[13]. Users can toggle ad personalization on or off at any time and delete ad-related data whenever they choose ^[14]. Ads are labeled, and OpenAI says they do not influence the model's responses ^[15]. Advertisers do not get direct access to conversations; OpenAI does the targeting internally using contextual signals without exposing raw dialogue ^[16]. OpenAI notified users of the changes by email, walking through the new controls ^[17].

For buyers, that last point is the structural constraint. A channel where the platform holds the signals and the advertiser sees none of the underlying dialogue is a black-box buy: measurement and audience definition depend entirely on what OpenAI chooses to expose ^[16]. The inventory is also concentrated at the bottom of the pricing ladder, since the paid upper tiers carry no ads ^[3], which shapes who is reachable regardless of consent rates.

What to watch: whether OpenAI publishes opt-out rates in the US and opt-in rates in the EU and UK, because the gap between them determines how much of the European base is addressable at all ^[8][9][11]. Watch for the consent treatment in Mexico, Brazil, Japan, and South Korea to be clarified ^[12]. And watch the labeling and no-influence commitment ^[15], which is the claim regulators and users are most likely to test now that the product is live in six markets rather than one ^[6].