Autocorrect has “fixed” plenty of texts nobody asked it to fix. From October 29, ChatGPT gets the same power over your ads, and you won’t see every edit first.

Your ads get a ghostwriter: OpenAI’s creative enhancements tailor your ChatGPT ads to the conversation and the user’s interests, using your ad and landing page.

On October 29, they switch on by default for new and existing campaigns. They can:

• Rewrite headlines and descriptions
• Translate your copy
• Resize images
• Send clicks to a different page on your site

Enhanced versions can run without your approval, and there’s no preview of every one.

Want the final say? Opt out in Ads Manager before October 29. Campaign settings beat the account default, so you can also test it on one campaign.

Spotted in the wild: Search Engine Watch caught ChatGPT testing two more ad features:

• Single-brand shopping ads at the bottom of answers. One site’s name and favicon sit above grey cards with product names and prices, so no competitor shares the unit.
• A CTA button inside ads, with a dotted underline and an arrow. ChatGPT ads had no CTA before, and a clear next step may earn more clicks.

Case closed on the missing tags: After the GPT-6 release, ChatGPT links stopped carrying the utm_source=chatgpt.com tag, which tells your analytics a visit came from ChatGPT.

ChatGPT links also hide referral data, so analytics tools counted many of those visits as direct traffic.

OpenAI confirmed it was a bug and started rolling out a fix on October 15.

What to do: Annotate October 8 to 15 in GA4 or Adobe Analytics, and leave your attribution models alone.

Your ads may change faces, but at least your ChatGPT clicks got their name tags back.