Google Rewrote Its Rules on AI Content and Fake Authors. Your Author Now Has to Exist.

On October 1, 2026, Google updated three documents at once: its guidance on generative AI content, its people-first content documentation, and its rules around author profiles. The direction is not subtle. Content needs an author who is a real, verifiable person — and “real” now means corroborated, not merely named.

This landed one week before the September 2026 spam update finished rolling out (September 24 to October 8, worldwide, all languages) — an update Google itself described as using AI systems to detect AI-generated spam. Seven ranking updates are confirmed on Google’s Search Status Dashboard for 2026 so far. The pattern across all of them is the same: fabricated signals are being hunted by machines that are getting good at it.

What “fake author” now means

The old playbook is dead. It went: generate the article, attach a plausible headshot from a stock library, write “Sarah is a content marketer with 10 years of experience,” publish at volume. Every part of that pattern is now a detectable artifact:

• An author with no trace anywhere else on the web
• An author bio that describes credentials no third party confirms
• Dozens of articles with identical voice under names that don’t exist
• Stock photos reverse-image-matchable to image libraries

Google’s updated guidance targets exactly this: authors who exist only as markup. The byline is no longer a formatting detail. It’s a claim — and claims need evidence.

A verifiable author is an entity, not a name

Think of your author the way you think about your company: an entity that should be described the same way everywhere it appears. Concretely:

On your site. An author page with a real photo, actual credentials, links to profiles that exist, and a first-person account of experience with the topic. Not a bio — a dossier.

On the open web. The same person, the same photo, the same description of what they do — on their LinkedIn, on guest posts, on conference pages, on directory and community profiles that accept author or team information. Independent pages describing your author consistently are the corroboration that turns “a name” into “a person.”

In the content itself. First-hand detail that can’t be generated: numbers from your own runs, screenshots of your own dashboards, opinions you’d defend, mistakes you actually made. This is the part of E-E-A-T that no amount of markup substitutes for — experience is only visible when it’s specific.

The consistency test

Here is a cheap audit that takes an hour. Take your three most important authors. For each, search the name in quotes plus your company. Count the independent pages that describe them consistently — same role, same photo, same specialty. Then count the contradictions.

Most teams discover their authors are one strong profile (their own site) and silence everywhere else. That’s the same gap that shows up in company-level entity data: your website says one thing, the rest of the web says nothing, so the model has a single unverified source to work with.

Where off-site author data comes from

Author corroboration mostly doesn’t come from press or viral posts. It comes from boring, structured, persistent places:

• Directory and platform profiles where your team is listed alongside what you build
• Community accounts with post history under the real name
• Podcast, webinar, and event pages carrying the bio verbatim
• Bylined guest content on sites that already get crawled

These are the pages assistants and search engines can both parse, and they’re the ones you can actually control the timing of. When we run submission campaigns at BackLink, keeping the product description and team description byte-consistent across every listing is treated as part of the deliverable — because a directory listing that contradicts your author page isn’t an asset, it’s noise.

What to do this month

1. Inventory your authors. Real name, real photo, real credentials. If any of them are partially fictional, decide now whether to credit real people or consolidate under one verified identity.

2. Rebuild author pages as evidence. Credentials, links to external profiles, first-hand artifacts in the articles. Link out to the places that describe the author — outbound links to corroborating profiles help systems connect the dots.

3. Push consistent descriptions outward. Every profile, listing, and bio should say the same thing: who this person is, what they work on, why they’re qualified. Divergence between your author page and your directory listings is now a liability.

4. Kill the volume pattern. Publishing velocity without author depth is the exact footprint the September spam update was built to find. Ten articles with real experience beat a hundred with none.

FAQ

Does this mean AI-assisted content is penalized?

No — Google’s guidance has consistently said the issue is content created for search engines, not content with AI involvement. The October update sharpens the author requirement, not the tooling. AI-assisted drafts under a verifiable human author who adds real experience are within the guidelines.

We have one founder writing everything. Is that a problem?

No — a single well-corroborated author is a strong setup. The problem is having zero corroborated authors, which is the default state for most sites. Depth beats breadth in author entities.

How much of this is about Google versus AI assistants?

Both. Assistants resolve entities the same way search engines do: by finding consistent descriptions across independent sources. Fixing author corroboration helps you in AI Overviews citations and in ChatGPT/Perplexity answers at the same time.

Is a LinkedIn profile enough for corroboration?

It’s a floor, not a ceiling. One profile is a single source. Multiple independent pages — community history, guest bylines, directory profiles — create the pattern that both search and assistants treat as evidence.

What if our authors want privacy?

Then use a pen name consistently and build the corroboration around it — the requirement is verifiability, not exposure. What doesn’t work is an author with no external footprint at all.