The failure of automated loopholes
Google added fabricating creator profiles to its helpful content documentation in October 2026. This policy classifies using AI-generated headshots, made-up names, or false credentials to make content look like it was written by human experts as a form of deception. This update follows the March 2024 scaled content abuse policy, which applies to content made through automation, human efforts, or a combination of both. In 2023, Sports Illustrated published product reviews under authors that did not exist, including one author whose headshot was for sale on a site selling AI-generated portraits. Google has fought "cookie cutter" approaches since its 2003 Webmaster Guidelines warned against little or no original content. In 2010, John Mueller told a site owner that publishing Google Translate output directly could violate guidelines because it constitutes auto-generated content. In 2012, Google published a help page that listed automated synonymizing and scraped feeds as spam. The 2024 scaled content abuse policy applies whether content is made through automation, human efforts, or some combination of human and automated processes, which targets the very tools C-level marketers now use to replace real experts. Leaked Google documentation from May 2024 showed that Google stores the author of a document and uses an attribute called isAuthor. Sites that lean on these approaches often end up worse off than if they never tried them, as they spend years cleaning up while traffic and visibility stay suppressed.
E-E-A-T and the cost of automation
I watched a client spend three years recovering from the 2018 Medic update after they scaled medical content using automation without expert credentials. Because that site fell into the YMYL category, it lacked the necessary E-A-T for medical advice. When the update hit, Danny Sullivan pointed site owners to the Search Quality Rater Guidelines, though his tweet has since been deleted. We eventually rebuilt the site around real doctors and medical reviewers, which allowed the site to regain its previous performance levels. My 2019 analysis of 64 sites showed that sites winning the 2018 and 2019 core updates were 258% more likely to use real experts on their content than those losing sites. Google mentions E-E-A-T 120 times in its Search Quality Rater Guidelines, and Bing uses language focusing on quality and credibility. I spent months in 2019 reading and re-reading these guidelines after the Medic update. I even printed and laminated a copy to keep on my desk. This experience proved that investing in real humans is the only strategy that survives these cycles. In 2021, I wrote an article specifically to debunk misconceptions about E-E-A-T being a checklist or a single ranking factor. In 2022, I gave a MozCon talk about focusing on real expertise to avoid being penalized or devalued.
Why AI cannot replicate human experience
Real humans provide pattern recognition from lived experience that a large language model cannot replicate. An LLM can summarize what others have written about a job, but it has never actually done the job. A 2026 study in PNAS Nexus found that leading AI models score about as well as people on creativity tests, but their answers are far more similar to each other than people’s answers are. Personal profiles on LinkedIn outperform company pages by 2, 4, or 8 times on impressions depending on the specific data used. People follow people, and the Cambridge Dictionary named "parasocial" as its 2025 Word of the Year to describe the connection users feel with online personalities they do not actually know. In 1966, researchers found that a highly competent person became more likable after spilling a cup of coffee, a phenomenon psychologists call the pratfall effect. A real expert leaves a trail of conference photos, videos, podcast interviews, and social media replies that AI cannot manufacture. This pattern recognition only comes from lived experience, and it is often the exact level of helpful insight that users look for when they search for advice. As AI makes everything else cheaper to produce, the value of an earned reputation keeps climbing. In nine experiments with more than 6,000 people, the same supportive message received higher ratings for being empathic and more genuine when participants believed a human wrote it. AI can imitate empathy and humor, but it cannot draw on experiences it never had. This makes AI writing feel stale. A real person brings something new to the conversation. Google’s helpful content guidance asks whether content provides "original information, reporting, research, or analysis," and this still requires a person to decide which questions are worth asking. An LLM can summarize what others have written, but it has never sat with a scared patient or lost a big client.

