Fake authorship and Google: what has actually changed

TL;DR

In October 2026 Google began calling fabricated author profiles, such as AI headshots, made-up names and false credentials, “a form of deception” in its site-owner guidance. It has not said how its systems detect and punish them.
The known exposures, from Sports Illustrated to “Dr Eleni Nicolaou”, came from newsrooms. There is no public case of Google penalising a site for fake authors alone.
For SEO and GEO work, invented experts are now a liability with no upside. Real, accountable people and documented human review are what will hold up.

Why this is back on the table

Fake authorship covers invented writer profiles, personas with synthetic headshots, non-existent experts quoted in press releases to win links, and real names or credentials used dishonestly. The aim is always to borrow trust the content has not earned.

Two things happened in October 2026. Google’s people-first content guidance gained a paragraph telling site owners to avoid deceptive authorship information, and within days OpenAI described Bogus Bylines, an Iran-origin operation whose seven fake Western journalists placed almost 100 articles in roughly a dozen outlets.

How we got here

The modern pattern was set in November 2023, when Futurism traced Sports Illustrated reviewers such as “Drew Ortiz” to a marketplace selling AI headshots. Since then the cases have grown in number and reach:

April 2025: Barbara Santini, one of the UK’s most quoted psychologists, turned out not to exist. Press Gazette’s wider dossier now runs to more than 1,000 faked stories, many placed to win links from trusted domains.
June 2025: VRT NWS found that “Sophie Vermeulen” had written 403 articles for Elle Belgium that year, with similar invented profiles at Marie Claire and Psychologies.
September 2026: “Dr Eleni Nicolaou”, an art therapist fronting a paint-by-numbers retailer, had appeared more than 30 times in outlets including Forbes and Vice before Qwoted removed her profile.

The common thread is who the byline was for. These personas were built for a ranking system people believed was checking who wrote things, not for the reader.

What Google has actually done

In January 2024 Google’s Search Liaison said bylines do not help pages rank, and the March 2024 spam policies targeted mass-produced content rather than the byline itself. And it didn’t stop there.

The rater guidelines went further. Since January 2025, section 4.5.3 has listed fake creator profiles, AI-generated author images and false expertise claims as deception deserving the Lowest rating. Raters do not set rankings, but they show what the systems are trained to recognise, so raters guidelines are considered to be a good benchmark to update one’s content against.

This year, the pace has been picking up. In May, Google’s spam policies were extended to AI Overviews and AI Mode, and in August the site reputation section was rewritten around four review factors, one of them “stated or implied authorship”. After talks with the European Commission, those manual actions no longer affect rankings for EEA users.

Between guidance and enforcement

Google has not explained how its systems would recognise a fabricated profile. The 2024 Content Warehouse leak showed author fields such as isAuthor, but nothing public shows how much weight they carry, so fake authors seem to be caught as a symptom of scaled content or site reputation abuse.

Enforcement is also getting harder to understand for an outsider. The September 2026 spam update ran for 13 days and 16 hours, against the one to two days which are considered to be the norm. Consultants reported more manual actions in October, mostly for thin or scaled content. None was documented as a fake-author action.

Where AI search changes the picture

AI answers drop the byline. A fake expert quoted in a national paper becomes “experts say” in a synthesised answer, carrying the outlet’s authority with no name to check, and AI engines already lean towards earned media over brand-owned pages, which is exactly where these personas were placed.

The citation layer is a weak filter. Washington University in St. Louis audited 98,020 AI Overview claims and found 11.0% unsupported or contradicted by the cited pages. Because retrieval happens at query time, a Reddit thread can surface as fact within days, and a planted expert quote can travel the same route.

What this means in practice

The sensible response is accountability rather than E-E-A-T theatre. Use a house byline for content no single person wrote instead of inventing a persona, and name a real subject-matter reviewer for AI-assisted work, with a record of what they checked.

Digital PR needs the same discipline in reverse. Only put forward spokespeople who will take a call and have a public expert profile a journalist can verify, and audit legacy author pages for stock headshots and unsourced credentials.

My forecast is cautious. Google will probably keep tightening its wording rather than launch a dedicated fake-author classifier, so the main risk is reputational: an exposed fake expert can damage the brand behind it for years. It is worth doing the check.