DEV Community

Breach Protocol
Breach Protocol

Posted on Originally published at groundtruth.day

OpenAI's ad policy now lets it refuse ads that threaten its 'competitive position', after Adobe's image-tool ads were cut from ChatGPT

OpenAI changed its advertising rules on 10 September 2026 so that it can decline ads that conflict with its "business interests, or competitive position," one day after The Information reported that ChatGPT had begun blocking ads for rival AI image and audio products. Adobe, which had advertised its Firefly image generator during ChatGPT's ad pilot, was among the companies caught by the change.

Key facts

  • Policy version 1.6, dated 10 September 2026, adds "business interests, or competitive position" as grounds to refuse ads.
  • When: The Information published its report on 9 September, and the policy page changed the following day.
  • Who: OpenAI; Adobe was among the advertisers affected, according to the report by Alix Coutures and Amir Efrati.
  • Primary source: OpenAI's ad policies page.

What OpenAI's rules now say

ChatGPT began showing ads this summer, a step Ground Truth covered when ads went live in beta. The rulebook for those ads is a public policy page, and its wording changed this week.

The current version reads: "OpenAI reserves the right to decline, restrict, or remove advertisers and to refuse to display ad content and website links for any reason, including where they conflict with our advertising principles, business interests, or competitive position." The page's changelog describes version 1.6 as "updated to clarify our right to decline ads where they conflict with our advertising principles, business interests, or competitive position."

The previous version, dated 10 August and preserved by the Internet Archive, said only that OpenAI "reserves the right to refuse to display ad content and website links for any reason." Legally, "for any reason" already covered it. What the new sentence adds is candour: competition is now a stated ground.

What advertisers were told

According to The Information, OpenAI is "blocking ads in ChatGPT for AI image and audio products that compete with its own features, catching advertisers such as Adobe by surprise," as the publication summarised on X. Search Engine Land reported that "the change was reportedly communicated directly to advertising partners rather than publicly announced," and that Adobe's Doug Wyatt reportedly said OpenAI told partners that campaigns for standalone image and voice generation products would no longer be approved. Video-generation tools could still advertise.

Investing.com added that Adobe had used ChatGPT's initial pilot to promote both Acrobat Studio and Firefly. It also reported that ChatGPT "has simultaneously stopped returning outbound citations and external links for action-oriented image queries," such as searches for photo editors.

Why it matters

The move is ordinary for a platform and unusual for an assistant. Retail marketplaces routinely favour their own products, and regulators have spent a decade arguing about it. A chatbot is different because it is not only a shelf. It also answers the question "which tool should I use?" When the same company sells ad space next to that answer, builds competing features, and now reserves the right to turn rivals away on competitive grounds, users have fewer ways to tell advice from promotion.

Think of a travel agent who also owns an airline, and who has just stopped taking brochures from other airlines. The advice might still be good, but you would want to know about the arrangement.

For developers building on OpenAI's platform, the signal is about categories. Image and audio generation are areas where OpenAI now ships its own products, and the report says those are exactly the categories that lost access to ChatGPT's ad inventory.

The caveat

OpenAI has not publicly explained the policy or confirmed which ad categories are restricted; the image and audio detail comes from The Information's reporting on what partners were told, and the public policy page names no category. The claim that ChatGPT stopped linking out for some image queries also comes from secondary reporting and has not been independently measured.


Originally published on Ground Truth, where every claim is checked against the primary source.

Top comments (0)