Sponsored by
Halfway there, folks.
Every content team running AI at volume has the same quiet agreement. Somebody will check it. Probably. Before it ships, more or less.
Google just wrote that agreement into its guidance.
On October 1 it added three sentences to its guidance on generative AI content. Models "don't retrieve facts, but predict a likely sequence of words," outputs may contain hallucinations, and then the line that matters: "It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing."
It isn't only the body copy, either. The review covers metadata, so the meta descriptions and alt text your team generated by the thousand last quarter are in scope too.
Ecommerce teams get the stricter version. The page points to Merchant Center rules: AI-generated product images must carry metadata saying so, and AI-written titles and descriptions have to be labeled as AI-generated. It also suggests telling readers how automation was used. Optional, for now.
Google's people-first content page, in a version dated October 1, now carries a paragraph on authors. Inventing creator profiles with AI headshots, made-up names or false credentials is "a form of deception" and a low-quality signal, SEJ reported. Google says any deception makes a page untrustworthy "to both users and our automated quality systems."
That one should worry more people than it will. If your blog has a "senior analyst" nobody in the building has ever met, Google now has a word for it.
To be precise about what changed: this is guidance, not an algorithm update, and Google says its raters don't directly decide rankings. But several well-known SEOs are predicting an update, and guidance is usually where Google fires the warning shot.
The real cost is time. A human fact-check on every AI draft isn't free, and most content-at-scale business cases quietly assumed it was. If your team ships 200 AI-assisted pages a month, somebody needs to own the check, the hours and the sign-off.
So go look at last month's output. Could you name the person who checked each page, and would you want Google to ask?
Today’s sponsor:
State of Product reveals what AI still hasn’t solved
80% of product professionals say AI helps them ship faster, but customers aren’t seeing value any sooner. Atlassian’s State of Product 2027 explores the gains AI is delivering and the challenges that remain, from decision-making that hasn’t kept pace to gut instinct overriding customer evidence.
TALK ON THE MARKET
Eric Seufert, the analyst behind Mobile Dev Memo, on whether shopping agents replace ads:
The agentic-commerce decks skip this. Auctions exist because every merchant values a customer differently.
A flat fee flattens that. Expect the ads to follow the agents, not the other way around.
Want to see case studies of AI-powered campaigns?
Try our sister newsletter:
NUMBERS MATTER
60% to 50%. Senior marketers confident they're balancing short- and long-term goals, one year apart, in a Kantar preview shown at The Drum Live. It's based on 800 senior marketers, and the full study isn't out yet.
More data, more dashboards, less confidence. AI is making the reporting faster, and nobody's more sure what it says.
DAILY PROMPT
Our Meta Title and Description Optimizer, because Google just put AI-written meta tags in the review queue:
Optimize meta titles and descriptions that lift click-through rates from search results while using target keywords naturally.
Optimize meta tags for the following:
**Page Information:**
- Page URL: [YOUR_PAGE_URL]
- Page Topic/Main Keyword: [PRIMARY_KEYWORD]
- Secondary Keywords: [SECONDARY_KEYWORDS]
- Target Audience: [TARGET_AUDIENCE]
- Current Meta Title (if applicable): [CURRENT_TITLE]
- Current Meta Description (if applicable): [CURRENT_DESCRIPTION]
- Page Content Summary: [BRIEF_DESCRIPTION_OF_PAGE_CONTENT]
- Brand Voice Notes: [BRAND_VOICE_NOTES]
Fetch the live page at the URL and read any files I attach (Search Console exports, keyword data, style guide) before writing. Then search the current results page for the primary keyword and note how the top listings are positioned, so my tags differentiate rather than echo them.
**Requirements:**
- Meta Title: 50-60 characters, verified by counting characters in code rather than estimating
- Meta Description: 150-160 characters, same verification
- Include primary keyword naturally in both title and description
- Lead with the value proposition that matters to the stated audience
- No keyword stuffing, no filler adjectives
- Match the intent signaled by the page content, not just the keyword
- Reflect the conventions visible in the live results for this query
Provide 3 alternative options for both the meta title and description. For each option, give the exact character counts, explain the positioning choice, name the search intent it targets, and state what you would measure to judge whether it beat the current tags.Use it on your top 20 pages by traffic, then have a person read every option before it ships. That last step is now in Google's documentation.
CMO CORNER
Time is not your enemy. Misuse is.
AI handed most content teams real hours back, and plenty spent them on more pages. Those hours belonged to the checking and thinking that make a page worth ranking.
We provide consulting services for CMOs on AI strategy for marketing teams through our Teamless offering.
TOOL UPGRADE
If Google wants a human check on every AI draft, give that human better tools:
Originality.ai — best for SEO and editorial teams screening freelancer and contributor drafts before they publish. Try Originality.ai here →
Paraphraser.io — best for editors tightening a clunky AI draft into plain English before it goes to review. Try Paraphraser.io here →
Holo AI — best for small teams that want on-brand ads, posts and emails drafted from their own website, ready for a person to check. Try Holo AI here →
→ We have 300+ curated, filtered, and categorized AI tools in our directory. Browse to find the best ones that help your marketing team.
LEARN THIS
For whoever just inherited the fact-check: AI Content Quality Assurance Framework. A review process that holds up at volume.
Steal the checkpoint system and name an owner for every step.
→ We publish workshops, courses, guides, manuals, and more on AI marketing for free. Browse our learning resources for free.
BRAND SLOP
Instinct, the personal AI agent that just raised $1 billion at a $10 billion valuation, didn't wait long before trying to sell its users things.
It started pushing unsolicited product picks, built from what it found in their email and upcoming trips. One founder wrote on X about "my first 'ewwww' moment with an AI assistant" when the suggestions started arriving, TechCrunch reported. An investor's verdict, which I'll paraphrase, was that it had turned into a bad Amazon-recommends app overnight.
Instinct says its goal is "world-class recommendations." Some users heard an ad that had read their inbox.
Every sponsored placement inside an agent will live on that line. Would your brand want to be the recommendation somebody screenshots with "ewwww"?