Google raises the bar on AI content as publishers start counting citations
Google wrote down 2 things it had only implied before: that generated copy needs a human factchecker, and exactly how its quality raters score the main content on a page. A publisher coalition shipped a standard for counting what AI systems do with your words, and agent checkout picked up another identity check.
This Week's Signals
Google now says to factcheck every AI-generated page before publishing
Why it matters for your score
If you publish generated copy at any volume, your cheapest protection is a named human who checked the facts and a date that proves when.
Google documents the 4 attributes raters use to judge main content
Why it matters for your score
Effort, originality, talent or skill and accuracy are now written down, so you can audit your own pages against the same list Google uses.
Publisher coalition SPUR ships its Content Telemetry standard at version 1.0
Why it matters for your score
Counting whether AI retrieved, grounded, cited, presented and engaged your content is becoming a standard, which turns citation gaps from a hunch into a number.
DataDome joins Experian Agent Trust to verify shopping agents mid-session
Why it matters for your score
Checkout is being rebuilt to tell an authorised agent from a hostile one, so blanket blocking of anything that is not a browser is getting more expensive.
ChatGPT adds virtual try-on and a Favorites library for shopping
Why it matters for your score
Product discovery keeps moving inside the assistant, so your images, variants and attributes have to be machine readable rather than just well designed.
Google Ads starts turning offers found on your site into ad copy on 12 October
Why it matters for your score
Google will read your landing pages and promote whatever deal it finds, switched on by default, so a stale offer on your own site becomes your advertising.
Emerging Audit Signals
Quality guidance is converging on human effort as the one input a model cannot fake.
Publishers are building measurement before they build blocking, because you cannot price access you cannot count.
Agent commerce is being solved at the identity layer well ahead of the discovery layer.
Platforms increasingly read your site themselves and publish what they find, with or without your input.
What to do this week
- 1
Put a named human reviewer on every AI-drafted page before it goes live, and log the date they checked it. Google now treats unreviewed generated text as a quality risk, and your defence is a record, not an intention.
- 2
Open your Google Ads account level automated assets settings this week and decide about automated promotions before 12 October. If you leave it on, every offer sitting on your landing pages is live ad copy you did not approve.
- 3
Take your top 20 pages and score each one on effort, originality, skill and accuracy, the 4 attributes Google just documented. Run /tools/ai-readiness-scan beside that list so you can see what agents can actually read on those same pages.
Sources
- ↗ Using generative AI content, Google Search Central
- ↗ Creating helpful, reliable, people-first content, Google Search Central
- ↗ Latest Google Search documentation updates, Google Search Central
- ↗ SPUR Content Telemetry Standard v.1 published, SPUR Coalition
- ↗ Content Telemetry: an open standard for reporting AI content usage
- ↗ Experian checks on AI shopping agents gain DataDome behavior monitoring
- ↗ ChatGPT can now virtually try on clothes for you
- ↗ Google Ads will automatically pull promotions from advertisers websites
- ↗ About automated promotions, Google Ads Help