Google Just Called Fake Author Profiles Deception. Here Is the 5-Step Byline Audit.

On October 1, Google rewrote its "Creating helpful, reliable, people-first content" guidance. Most coverage focused on the new definition of main content; the sharper paragraph sat in the "Who (created the content)" section, right after the existing byline advice. Google now says fabricating creator profiles, using AI-generated headshots, made-up names, or false credentials to make content look written by human experts, is "a form of deception" and "a signal of a low-quality page." Search Engine Journal documented the change here.

Back in January 2024, Google's Search Liaison said bylines do not help pages rank. That is still the position. A real byline buys you nothing in the rankings. But as of October 1, a fake one is documented as a trust problem for Google's automated quality systems. The upside of author markup stayed at zero. The downside moved.

Why this is happening now

Google engineers at the Search Central Live Deep Dive in Barcelona said scaled content is now a bigger problem than link spam, and that spam updates use AI to catch AI-generated spam. John Campbell's recap of the event has the details. The same week, Google added a second AI-content rule to the same guidance: generative models predict word sequences instead of retrieving facts, so fact-check AI-assisted content by hand before publishing. SEJ covered both changes together.

Then Lily Ray, the closing speaker at SEO IRL in Toronto today, posted that all signs point to a major Google update on the horizon, pointing directly at these documentation edits. Read that sequence the way an SEO reads anything Google publishes twice: the docs are the warning shot. Google's September spam update is in its final rollout phase. The daily SER recap has the timing.

Two honesty notes before the playbook. First, the new paragraph does not explain how automated systems would identify a fabricated profile, and SEJ's reporting is explicit about that gap. Second, rater guidelines are not the ranking algorithm. Section 4.5.3 of the rater guidelines already rated deceptive author profiles as Lowest quality, but raters do not set rankings. Treat this as Google's stated quality bar, not a ranking formula. The four attributes Google named for main content, effort, originality, talent/skill, and accuracy, describe what quality looks like. Marie Haynes flagged the change and Sunny Patel broke down the four attributes.

The 5-step byline audit

Step 1. The name test

Every page should attribute content to a real human, spelled identically everywhere: the visible byline, the schema author name, the about page, and the linked social profiles. If your schema says one name and your byline says another, the page looks assembled, not authored.

Step 2. The photo test

The author photo must be a real, current photo of that human. Not a stock headshot, not an AI portrait. Google's new paragraph names AI-generated headshots explicitly. A reverse-image search should find the same face attached to the same name elsewhere, not the same face attached to five different "contributors" across the web.

Step 3. The credential test

Every title and credential in the author bio must survive thirty seconds of verification. Employer, role, certifications, "featured in" claims. If a claim cannot be confirmed by a second source you do not control, delete it or prove it. False credentials are named in the new paragraph right next to fake names. Inflated titles are the most common way honest sites trip this wire.

Step 4. The sameAs test

Your structured identity must be identical across every template on the site. Same Person @id, same name spelling, same sameAs URLs in the homepage schema, the about page, and every article. Then click each sameAs URL and confirm it resolves to a real profile for the same person. Machines do not forgive two different LinkedIn URLs for one author. I know, because I failed this one. More on that below.

Step 5. The four-attribute test

Read your own main content against Google's four attributes: effort, originality, talent or skill, accuracy. Would a skeptical reader conclude a person did the work? Original reporting, real data, named processes, and claims a reader could check are what those words cash out to in practice. This is also where getting cited by AI answers and author trust meet: systems that quote you are implicitly vouching for the name attached. And if any draft was AI-assisted, the new guidance is blunt. Manually fact-check it before publishing, and document that process somewhere a reader can see it.

I ran this audit on my own site

I ran all five steps on dalindegraff.com while writing this post. The wins: my real name is spelled identically in every byline, every schema block, and my about page. My author bio links to profiles I actually control. The miss: my own site pointed at two different LinkedIn URLs for one author. The Person schema used one URL and the article schema on blog posts used another, and neither matched the profile I actually use. One author, two URLs on my own site, neither of them mine. That is exactly the kind of inconsistency the new guidance trains machines to distrust. I pointed every template at my real profile before this post went live.

Google previously said it was not worried about faked authorship. Barry Schwartz noted the shift. Times changed. The warning arrived as a documentation edit, which is often how Google tells you what is coming next.

When did you last click the author link on your own site and check where it leads?