On October 30, every live ChatGPT Ads campaign started running creative its owner never approved. OpenAI calls the feature creative enhancements. It switched on by default for any account that hadn't turned it off.
If you run ads on ChatGPT and left your settings alone that week, OpenAI's system can now rewrite your headline, translate your copy, resize your image and send the click to a different page than the one you picked.
What flipped on October 30
The notice went out on October 20. OpenAI emailed advertisers that creative enhancements would switch on in about ten days unless they acted first. The email didn't hedge.
It told advertisers, "You can opt out of creative enhancements before October 30, 2026 or change your settings at any time in Ads Manager." Nine days later, according to OpenAI's own ads help page, the default flipped for new campaigns and every existing one that hadn't said no.
The same week brought two more changes. A multi-image ad format opened to every advertiser, after testing since October 26. It shows four to six square images behind one destination URL and a single "Learn more" button.
Search Engine Roundtable also reported a wider desktop ad unit. It now fills more of the conversation column, applied on its own with no setting to flip. None of it needed an advertiser to do anything. Across all three changes, the default itself moved, and the job shifted from choosing a setting to noticing one.
What creative enhancements can touch
OpenAI's own description of the feature is almost cheerful. It says the goal is making ads "more relevant to the conversation and/or user interests, helping improve performance, while reducing the work of creating multiple versions." The mechanics behind that line reach further than most advertisers expected.
It can write a new headline or description to match whatever the person is asking ChatGPT about. It can translate your copy into another language without telling you first. It can resize an image to fit a format you never built for. And it can point your destination URL at a different page on your own site, whichever one its model picks.
None of those versions gets a preview before it runs. Search Engine Journal flagged a gap OpenAI's own docs don't close. Does a UTM tag on your original link survive when the destination changes underneath it? Nobody outside OpenAI knows yet.
Why OpenAI is pushing this now
ChatGPT Ads is still young by ad-platform standards. OpenAI staged the whole rollout carefully, publishing ad principles months before the first ad ran. The program expanded to 26 more countries on October 21, taking it to 89 total. OpenAI says over 100,000 advertisers have run a campaign since testing began.
Growing that fast brings two problems. A lot of new advertisers write thin, one-size-fits-all copy, and the sales team feels pressure to prove the inventory earns its keep before the next round of country launches. Automating the creative is the obvious fix for both. It lets OpenAI's own model patch a weak headline instead of waiting for the advertiser to notice a flat click rate.
The multi-image format fits the same push. Four to six images behind one link and one button suit mobile scrolling inside a chat window, not a traditional display grid. Rolling it out the same week as creative enhancements says OpenAI wants its system tuning copy and visuals together, not handing advertisers two separate tests to run on their own.
Advertisers are not thrilled
The reaction, once marketers noticed, ran from skeptical to furious. Anthony Higman put it bluntly to ppc.land: "If an ad platform is framing something as 'less work for you' you can pretty much guarantee what they really mean is: we are scamming you."
Melissa Mackey, who had already paused her agency's ChatGPT Ads test over lead quality, said she was glad she'd pulled back before this landed.
Know exactly what AI says about your competitors.
RankControl's Recon Agent monitors competitor citations across ChatGPT, Perplexity, Claude, Gemini, Grok, Google AI Mode, and Google AI Overviews. See where they show up and you don't.
The timing lands badly for a channel that was already short on trust. One marketer running campaigns for a big B2B client and a regional ISP called ChatGPT Ads traffic hard to scale, and about on par with Reddit for quality, even before this change shipped.
Another, on the same thread, surveyed 160 people and got zero human clicks on a ChatGPT ad. A system that now edits the creative behind those clicks, with no log of which version ran, doesn't close that trust gap. It widens it.
How to check your own settings
The fix, if you want one, takes a few clicks. Open Ads Manager and look for creative enhancements under account or campaign settings. A campaign-level choice beats the account default.
Leave the feature on for low-risk test campaigns, and turn it off for anything with brand copy or a landing page you've already tuned for conversion. If you'd opted out before October 30, that choice held. Nobody got quietly re-enrolled.
What you still can't do is see each generated version after the fact, or confirm your tracking survives a swapped destination URL. Until OpenAI writes that down somewhere, the safest move for a campaign you care about is to opt out and test the feature on your own schedule, where you're watching closely.
The bigger pattern behind it
This is the second time in two weeks that a major AI platform changed how ads behave without asking the people paying for them. Google did something similar when it expanded AI Mode ads to ten more countries. Existing Search and Shopping campaigns got enrolled on their own, with no opt-out at all.
Paid placements on both platforms are starting to act less like something you set once and more like something you have to keep checking.
26 content formats. Published on your domain. Matched to your brand.
Guides, comparisons, listicles, case studies, and more. RankControl generates content that gets cited by ChatGPT, Perplexity, Claude, Gemini, Grok, Google AI Mode, and Google AI Overviews.
That's a cleaner argument for the free side of AI search than the paid one. A citation ChatGPT gives your brand in an answer doesn't get auto-translated or rerouted to a page you didn't choose, since there's no creative to edit.
RankControl tracks how often your brand actually gets named across all seven AI engines, Google included, every week. You can see whether your free description is drifting before you ever spend money fixing the paid one. Keep an eye on Ads Manager yourself every time OpenAI flips another default, or let that weekly check run in the background while you decide what, if anything, is worth paying for on ChatGPT.
What to do before your next ChatGPT Ads invoice
Check your Ads Manager settings today, not after the next bill lands. If creative enhancements are on and you didn't put them there, decide on purpose whether you want OpenAI editing your copy. Don't find out from a changed conversion rate instead.
And whatever you decide about the paid side, keep tracking what ChatGPT already says about your brand for free. That number moves on its own schedule, with or without an ad budget behind it.
Your competitors are getting cited by AI. You're not.
Every day without citation tracking is a day your competitors pull ahead in ChatGPT, Perplexity, and Claude.
