Google Now Shows Which Product Titles Its AI Wrote Over Yours
A new Product titles report shows the version Google AI served next to the one you uploaded. Days earlier, an agency audit found competitor names in AI-written copy on a client's account.
Yash Mandlesha was auditing a client’s Google Ads account on August 23 when he found the ads talking about other stores. The copy mentioned “Myntra sales,” “Amazon Sales” and “Flipkart Sales.” None of those were the client. Google’s AI had written the lines.
Three days later a new report started appearing in Google Ads accounts. It shows which product titles Google’s AI wrote, and which ones the advertiser uploaded. Search Engine Land reported the rollout on August 26, and Mandlesha, a Google Ads consultant at the Indian agency Mediagram, described the feature on LinkedIn the same day:
“Google AI can now generate customised product titles and show them in your ads when it believes the AI-generated title is more relevant.”
The order of those two events is the story. Google has been writing product copy on advertisers’ behalf for a while. The screen that tells you which words were yours showed up this week.
What the report actually is
The Product titles report sits inside AI Max for Shopping campaigns, a beta feature set Google says uses its AI to improve Shopping performance. Switch on the setting called text customization and, in Google’s own words, its AI “can dynamically update your product titles to align with customer search queries and intent.”
Google’s documentation supplies the example. A feed entry reading “Mens Shoe SKU 123” can go out as “Men’s Waterproof Hiking Boot.”
The report puts both versions side by side with impressions, product clicks, click-through rate, cost and average cost per click attached. It joins the asset, landing pages and search terms reports in AI Max for Shopping’s reporting set.
The scope deserves a caveat, because the two accounts of it don’t line up. Search Engine Land says the report is “beginning to appear in Google Ads” without naming a campaign type. Google’s own help pages put it inside the AI Max beta. If you’ve never switched AI Max on, this isn’t describing your account.
The mechanism
Text customization doesn’t edit your title. It competes with it. Google builds an alternative from your Merchant Center feed, predicts which version wins against a given search, and serves that one. Your title becomes a candidate rather than the thing that runs.
That’s also how competitor names get into your ads. The system chases query relevance, and shoppers in India search for sale events by the retailer running them. Mandlesha’s fix list ran three items: clear text guidelines, strong brand controls, regular monitoring.
The first of those is a real Google product, and it’s been available since February. Rushil Grover, a group product manager on Google Ads, announced the global beta expansion on February 26, describing rules an advertiser writes in plain language:
“define specific terms to exclude or concepts to avoid, in your own words, with rules like ‘don’t imply our products are cheap’ or ‘don’t use language like ‘only for.””
The guardrail existed six months before the audit that needed it. It’s a separate beta you have to know about and turn on, and it governs the text the AI writes, not which searches your ads answer.
What this costs you
A product title is the shortest piece of brand copy most companies own. It’s what a shopper reads before anything else, and as we wrote in July, it’s also what an AI shopping agent reads instead of your ad campaign. Hand it to the machine and the sentence introducing your product to both a human and a bot gets written by a third party with a different goal.
The liability didn’t move with it. Google’s revised advertiser terms took effect July 1 with no acceptance step, and the text customization FAQ shows where responsibility sits. It explains how to opt out of automatically created assets. It offers no accuracy guarantee for the ones you leave on, and it documents no way to exclude a brand term from generated copy. That control lives in the separate text guidelines beta.
The case for leaving it on
The honest counter-argument is Google’s, and it’s a good one. A customized title only serves when the system predicts it beats yours, so the version that runs is, on Google’s own measurement, likelier to earn the click. Insisting on your original wording has a price. The new report is the first time you can see that price in cost per click.
Google published a customer number alongside the text guidelines launch: BYD got 24% more leads at 26% lower cost while using the guidelines to hold its brand standards. That’s the intended shape of this. Constrain the machine, then let it write.
It only works when the constraints are on. Mandlesha’s client had the rewriting running and the guardrail off, which is the default state of anyone who enabled AI Max before February and hasn’t revisited it since.
The verdict
Google shipped the rewriting first and the receipt second, and the six months between the guardrail and the report is where the damage lives. If you run Shopping campaigns with AI Max on, open the Product titles report this week and read what your products have been called. Not because the AI titles are worse. Because until now you were buying clicks on sentences you’d never read.
Quoted in this story
- Yash Mandlesha, Google Ads Consultant, Mediagram (source)
- Rushil Grover, Group Product Manager, Google Ads, Google (source)
Want your perspective in coverage like this? Get quoted.
Sources
- Search Engine Land: Google Ads adds reporting for AI-generated product titles
- Google Ads Help: About AI Max for Shopping campaigns (beta)
- Google Ads Help: Text customization frequently asked questions
- Google (The Keyword): We're expanding beta access to text guidelines for all advertisers globally in AI Max
- Google Ads Help: About reporting in AI Max for Shopping campaigns
- LinkedIn: Yash Mandlesha on Google AI product titles and AI Max text customization
This story is part of our running coverage: the full picture →
Get Net Effect.
The net effect of AI on your marketing: the stories that matter, twice a week, in five minutes.


