ChatGPT Ads Can Run Rewritten Copy Without Your Review

OpenAI’s planned creative-enhancement default gives ChatGPT advertisers a reason to check who can change an approved campaign. The feature can alter an ad’s wording and send a visitor to a different page on the advertiser’s site, without a separate approval for each version.

The company describes those powers in its creative-enhancements documentation. An advertiser email published by Search Engine Roundtable on October 20 announced automatic enablement for new and existing self-serve campaigns. The rollout date needs clarification. The email names October 30. The English help page we fetched on October 23 names October 31 and pairs it with the wrong weekday.

Marketers can make the account decision now. They shouldn’t have to resolve a contradictory calendar to decide whether their approved words can change.

What permission covers

The setting gives the platform room to adapt a supplied advertisement to a person’s conversation or interests. That includes rewritten text, translation, image resizing, and selection of another relevant destination on the advertiser’s website. A version can be made ahead of delivery or during the conversation. The documentation says the interface doesn’t preview every variation.

OpenAI’s help page states:

Campaign-specific settings take priority over your account default.

That hierarchy matters to an agency managing several campaigns. A person who checks only the account setting hasn’t established what every campaign permits. The page says existing opt-outs remain in place, with controls available through Ads Manager’s Manage settings.

We’ve covered an earlier default change involving ChatGPT’s use of customer form data. This change concerns the advertisement a customer encounters. A team might approve a particular claim for one audience and find that it has delegated more freedom over its expression than the approver understood.

A practitioner is already questioning the test

Melissa Mackey, director of paid search at Compound Growth Marketing, reacted in an October 20 post reproduced by Search Engine Roundtable:

We ended up pausing our ChatGPT ads test due to poor lead quality and now I’m glad we did.

That’s one practitioner’s account of a test, without a published sample or results table. It doesn’t demonstrate that enhancements hurt lead quality. The pause preceded the announced rollout, and the comment shouldn’t be used to invent a causal connection.

It does raise a useful question for a small B2B team. If the existing test isn’t producing suitable prospects, adding creative variation creates another explanation to investigate. More clicks won’t resolve whether the campaign is reaching people sales can help.

Mackey’s professional biography confirms their paid-search role. Their reaction is an operator’s caution, separate from the platform’s claim that adapting the ad will improve relevance.

The approved input still needs an owner

The Ad Tools Terms published October 21 address creative tools more broadly than this single setting. They place responsibility for campaign settings and applied generated creatives on the advertiser. They also disclaim guarantees that generated work will preserve the advertiser’s intended meaning.

That creates an approval problem a marketing manager can recognize. Imagine, as a hypothetical, that a team approves a software offer for a particular customer segment. Changing the destination to another product page can put a visitor in front of a different offer. A reviewer needs to understand that possibility when allowing automatic adaptation, even if every original asset was correct.

For a founder who signs off on campaigns between other work, the practical question is who holds that permission. A contractor can be responsible for setup while a product marketer remains responsible for the claims. Those people need the same understanding of which changes are allowed.

There is also a reasonable case for enabling the feature. Adapting existing material to a conversation offers a way to create variations without manually producing each one. Whether that helps a particular account is a testable business question. The documents don’t supply a measured improvement that this article can promise.

A team that needs approval of every delivered version has a clear reason to opt out. A team willing to delegate variation needs an agreed boundary and a way to assess the leads it receives.

Keep the rollout uncertainty visible

The October 20 email gives Friday, October 30. The English help page currently says “Thursday, October 31, 2026,” although October 31 falls on Saturday. The current German help page, labeled as machine-translated, gives October 29. We couldn’t establish one reliable activation date from those conflicting statements, and haven’t verified any advertiser’s live account setting.

The immediate work is to record the account choice and check campaign exceptions. Save the supplied ad and destination so there is a clear starting point for the test. Then judge the campaign on outcomes the business can inspect, including whether the resulting leads qualify. As with ChatGPT’s earlier budget changes, someone needs to own the setting after launch.

For teams following AI’s effect on marketing budgets, the concrete decision is whether the platform has permission to change a campaign that already passed review. Make that choice deliberately, and record who can change it next.