#What people are saying
Barry Schwartz over at Search Engine Roundtable pointed out that Google almost never uses the word “critical” in its docs. Search Engine Journal and Search Engine Land both flagged the same thing: the fact-check now includes the metadata that appears in results, not just the page body.
On X, the practical reaction was even clearer. One post put it this way:
Google basically just called out the “generate and ship” pipeline by name.
That’s still how a lot of agencies and tools work.
#Why this hits dealerships harder
Dealership websites already have an accuracy problem.
Model pages with last year’s MSRP. Service pages promising loaners or hours the store doesn’t actually offer. Specials that expired months ago. Inventory descriptions that don’t match the car on the lot. Finance claims a compliance person would never sign off on.
A lot of that content was already wrong when a human template produced it. Now a lot of agencies are just running those same templates through AI and pushing the output across dozens or hundreds of dealer sites.
That’s what makes it twice as bad.
Google sees the same inaccurate copy repeated across a bunch of domains. It sees pages where you can swap the dealer name and the content still works. That’s basically the definition of scaled content abuse—lots of pages made mainly to rank, with almost nothing real added for the person reading them.
One wrong fact on one site is a trust problem. The same wrong fact, written the same way, on a hundred dealer sites is a pattern. And patterns are what Google’s systems look for.
#What “check it” actually means
Google didn’t give a checklist. The useful version is short:
- Read every claim a customer could actually act on (price, payment, incentive end date, warranty, hours, shuttle, loaner).
- Check the fields Google can quote: title tags, meta descriptions, schema, alt text.
- Compare the page to other rooftops and to the vendor’s other clients. If swapping the dealer name still works, it’s a template.
- Ask who actually looked at it. “The tool wrote it and we published the batch” is exactly the workflow Google just called out.
Running an AI audit of the site doesn’t fix this either. The model is the thing that needs a human, not the thing you should hand the audit to.
#Bottom line
Go check your content. Ask your favorite AI if it’s accurate.
I guarantee you it’s not.
If you’re sick of vendors who are just using the generate-and-ship lazy AI model, reach out to Carbide Digital. We have some of the most accurate content on the internet.