ChatGPT ads can now rewrite themselves—what can advertisers control?
OpenAI says campaign enhancements can rewrite headlines and descriptions, translate text, resize images and optimize destination URLs using conversation context. They are now enabled automatically for new and existing campaigns, while advertisers cannot preview every variation and no independent accuracy or brand-safety result is published.
The short version
- What happened
- From 30 October, OpenAI says creative enhancements are enabled automatically for new and existing ChatGPT ad campaigns unless an advertiser has opted out.
- Who it affects
- A reader may see ad wording adapted to the conversation rather than copy approved line by line by the advertiser. For marketing teams, that shifts work toward maintaining accurate landing pages, setting campaign-level controls and auditing generated variants after they run.
- What remains unproven
- Official guidance updated 28 October; activation begins 30 October.
Keep in mind
Official guidance updated 28 October; activation begins 30 October.
No denominator for generated or reviewed variations.
No independent translation, destination or brand-safety evaluation.
Campaign metrics do not isolate the causal effect of enhancement.
There is no current-active advertiser, spend or campaign-outcome denominator.
The weekly-user figure is company-reported, not audited unique people.
No independent privacy, labelling or answer-influence audit is cited.
OpenAI is both platform operator and advertising seller.
Source analysis · 2 linked sources. Primary analysed source date: 21 October 2026. Latest linked supporting source: 28 October 2026.
Read the full report · 3 min The evidence, methods, implications and limitations
Advertisers now delegate more than placement
OpenAI's updated guidance says ChatGPT ad campaigns can now generate or revise headlines and descriptions, translate copy, resize images and optimize destination URLs. The system can use the context of a conversation or a user's interests alongside the original ad and landing page. Variations may be prepared in advance or created in real time, and they can run without the advertiser reviewing every version.[2]
Automatic activation changes who checks the final claim
Creative enhancements are enabled automatically for new and existing campaigns from 30 October, although existing account- and campaign-level opt-outs remain in place and advertisers can change the setting in Ads Manager. That makes current landing-page accuracy more consequential: a stale price, availability statement or promotion can become input to several generated variants. The official page advises advertisers to keep those details current, but it reports no pre-publication accuracy test, human-review rate or appeal denominator.[2]
The distinction that matters
What can change before an ad reaches a reader
The advertiser supplies the offer; OpenAI may adapt how it is presented to a particular context.
Source material
Ad plus landing page
Prices and availability need to be current
Enhancement
Rewrite, translate, resize or optimize URL
A variation may run without individual approval
Evidence
Ads Manager reports campaign metrics
No independent accuracy or causal performance test
The wider advertising system still lacks outcome evidence
The original 21 October announcement said ChatGPT ads had expanded to 89 countries and that more than 100,000 advertisers had run campaigns. That cumulative company-reported figure still lacks a current-active, spend, repeat-use or performance denominator. OpenAI says ads remain labelled and separate from answers, and its guidance says creative changes do not alter ChatGPT's answers. Neither source provides an independent audit of answer separation, translation accuracy, destination choice, consumer complaints or campaign results.[1][2]