The Maze: Businesses buying ChatGPT ads through OpenAI’s self-service campaign tool face a new default: the platform can change their messages without the advertiser approving each version. Advertisers can opt out. OpenAI’s help page dates the switch to October 29, 2026; its reported advertiser email gives October 30.
Source: OpenAI. OpenAI’s illustrative grocery-ad example changes the wording to fit weeknight-dinner planning. This is an example, not measured performance.
The Details:
The platform can edit what the shopper sees. ChatGPT ads already sit below answers and contain a headline, description, image and website link. OpenAI’s “creative enhancements” can rewrite the words, translate them, resize images or send a click to another relevant page on the advertiser’s website. They draw on the existing ad and its landing page—the page opened after a click—alongside the conversation or user interests. That extends the platform’s role from choosing an ad to changing its presentation. OpenAI’s grocery example connects an existing delivery offer to weeknight-dinner planning; it is an illustration, not a measured sales result.
Advertisers keep an opt-out, rather than a sign-off on every version. The official settings guidance says the change covers new and existing campaigns. Existing account-level and campaign-level opt-outs remain intact. A campaign’s own setting overrides the account default. Advertisers can change the option in Ads Manager, OpenAI’s campaign-management tool, before October 29 or later. The one-day discrepancy with the email remains unresolved, so October 30 should not be treated as a guaranteed extra day. The sources do not establish that every country or partner buying channel is included.
Less writing work is the pitch; better results are not established. OpenAI says adaptation should improve relevance and reduce the effort of creating multiple messages. But advertisers cannot preview every version. Reporting shows overall delivery and results, rather than proving which editing choice caused a gain. OpenAI’s separate ad-creation guidance recommends multiple distinct messages and coherent destinations. Its metadata-based suggested drafts still allow advertiser review and editing before assignment. That reviewed-draft process is different from letting enhanced ads run automatically.
MarketMaze Take: The important change is where the advertiser exercises control. With enhancements enabled, approving the uploaded message no longer means approving every message delivered. The decision becomes whether to delegate adaptation—and whether the website contains reliable material for the platform to reuse. This is an operational trade-off, not proof that automation will damage or improve sales.
The website becomes part of message control. A landing page already explains the offer to shoppers. Here it also supplies material used to adapt the ad. An old promotion or unclear product description can therefore matter before the shopper reaches the site. That dependency is useful for advertisers anywhere to understand, even where this feature is unavailable. OpenAI’s policies still require truthful ads and consistent destinations; automated rewriting does not remove that standard.
Policy review and brand approval answer different questions. OpenAI describes automated compliance checks with human oversight. That is a meaningful safeguard, but it does not give a brand a preview of every translation or headline. Keeping enhancements off preserves a tighter approval boundary; leaving them on accepts more platform discretion. Neither choice comes with a demonstrated performance advantage in the inspected evidence.
