Short answer
On 1 October 2026 Google updated two guidance documents: fabricated author profiles are now described as a form of deception, and hand fact-checking AI output is called critical. These are documentation changes, not ranking releases. But documentation is where enforcement starts.
Nothing ranked differently on 1 October. That is why most people ignored it.
But every spam policy Google now enforces appeared in documentation first. Scaled content abuse was written down before anything was demoted for it.
Change one: fake authors are now deception
The people-first content guidance gained a section on main content, and it calls fabricated author profiles a form of deception.
That wording matters. Deception sits in the same category as cloaking and hacked content, not in the softer language used for thin or unhelpful pages.
This is common and most sites doing it do not think of it as deceptive.
- Invented bylines with stock photographs and a plausible biography.
- Credentials nobody holds. A medical reviewer who is not a clinician.
- Composite personas standing in for a content team.
- Author pages with no verifiable existence anywhere else on the web.
The fix is not to remove authorship. It is to make it true. A named person who actually works there, with a profile that exists elsewhere, beats an impressive invented expert.
Change two: fact-check AI output by hand
The generative AI content guidance now calls it critical to fact-check AI output manually before publishing.
Google’s position on AI has not changed. How content is produced still does not matter. What this adds is an explicit expectation about review.
That closes a gap people have been using. The argument was that if AI content is allowed, volume is a business decision. This makes clear the allowance is conditional on human review of accuracy.
It also lines up with what moves AI impressions. An unverified figure is a liability to cite, which is covered in generative AI impressions dropped in Search Console.
Change three: Google named a recovery timeline
On 2 October, Google published timing guidance stating recovery after a core update typically takes three to six months.
This is the most useful of the three, because it is a number you can put in front of a stakeholder. It comes from Google, not from someone managing expectations.
It also quietly contradicts a lot of what gets sold. Anyone promising recovery in 30 days is promising something Google’s own documentation does not support.
What to actually do
- Audit every byline. Does this person exist, work here, and hold the credentials claimed? Remove or replace anything that fails.
- Check author pages resolve to real profiles elsewhere. LinkedIn, a professional register, published work.
- Add a review step that checks facts, not just grammar, and record who did it.
- Source your figures. Every statistic linked to where it came from.
- Reset timelines using Google’s own three to six months rather than a hopeful one.
Step one is the urgent one if you run YMYL content. Health, finance and legal sites with invented reviewers are now described by Google’s own words as deceptive.
Why documentation changes are worth watching
Because the gap between writing a policy and enforcing it is where the cheap fixes live.
Site reputation abuse was documented in March 2024 and enforcement began on 5 May 2024. Publishers who unwound their arrangements in that window were fine. Those who waited were not.
Nothing has been announced about enforcing the author guidance. That is exactly why now is the cheap time to fix it.
Education sites are particularly exposed to the author guidance, covered in education and course site SEO recovery.
Back button hijacking followed the same two-month pattern from documentation to enforcement, covered in back button hijacking.
Audit your authorship in an afternoon
- List every author shown on the site, including reviewers and contributors.
- Search each name outside your own site. A real professional usually has a LinkedIn profile, an entry on a professional register, or published work elsewhere.
- Reverse image search the author photos. Stock photographs are the quickest giveaway of an invented profile.
- Check every credential claimed against the body that issues it, where one exists.
- Decide author by author: keep, correct, or replace with a real person who actually reviews the content.
Replacing an author is not a risk in itself. Moving bylines to real people is exactly what the guidance asks for. Keep a short note of each change so the history is clear if anyone asks.
On YMYL sites, do this before anything else. A medical or financial reviewer who cannot be verified is the profile most likely to be treated as deceptive, and the one that carries the most risk for readers.
What the change does not mean
- It is not a ban on AI. Using AI tools remains acceptable when the output is checked and genuinely useful.
- It does not require an author box on every page. It targets invented people, not missing ones.
- It does not change existing rankings by itself. Guidance describes what systems look for; enforcement comes through updates.
- It does not say when enforcement will tighten. Nobody outside Google knows that.
Read it as a clear statement of direction. Sites that fix authorship now are ahead whenever enforcement arrives.
Not sure whether your authorship would survive scrutiny?
Invented bylines are now described by Google as deception, and the sites most exposed are the ones that never saw it that way. Send your Search Console access and within 48 hours I will tell you where you stand.
Get free recovery audit