Google Now Requires Manual Fact-Checks of AI Content

Google added a manual fact-checking requirement for AI-generated content to its Search guidance on Oct. 1, warning that generative models produce output that "may contain inaccuracies (also known as hallucinations)."

The update revises Google's guidance page for sites using generative AI. According to the company's documentation updates page, the change brings the guide in line with what Google has presented at developer events. The Oct. 1 changelog entry states the page was updated "with information from the Search Quality Raters guidelines." It is the only page listed for that date.

What exactly did Google add?

Three sentences now sit in the "Focus on accuracy, quality, and relevance" section, placed directly after the section's opening line. Compared with the Dec. 10, 2025, version, Google inserted this wording:

"Keep in mind that generative models don't retrieve facts, but predict a likely sequence of words based on their training data. Because of this, generative AI outputs may contain inaccuracies (also known as hallucinations). It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing."

The metadata line changed as well. "This includes metadata" became "This review also applies to metadata." The page still lists title elements, meta description elements, structured data, and alternate texts for images, "which can appear in Search results."

What does this change for content teams?

The Dec. 10, 2025, version of the guidance asked for accuracy but contained no manual fact-check instruction. The Oct. 1 update adds that step and supplies Google's reasoning for it.

The practical consequence reaches beyond article copy. AI-generated metadata, such as title tags, now falls under the same manual review as drafts. A review process built only around body text misses part of what the guidance covers. Teams using AI to write titles or image alt text must bring those outputs into the same workflow.

This is not a new position for Google staff. In an April 2024 episode of the Search Off the Record podcast, Gary Illyes said generative AI output always has to be fact-checked. In an August 2025 interview, Illyes said AI content "should be human curated," with someone verifying accuracy, as Roger Montti reported.

What stays the same?

The rest of the guidance page matches the Dec. 10, 2025, version. It still opens by saying generative AI can help with research and with adding structure to original content. The same paragraph warns that using AI tools to produce many pages without adding value for users may violate Google's spam policy on scaled content abuse.

The page also still points to sections 4.6.5 and 4.6.6 of the Search Quality Raters guidelines, which cover scaled content abuse and content made with little effort, originality, or added value. It notes that raters use those guidelines to evaluate Google's ranking systems and that their ratings "don't directly influence ranking."

The road ahead

Teams that use AI in content production can now cite Google's own wording when setting a review policy that covers both drafts and metadata. Google logs major changes to its Search documentation on the documentation updates page, and that page is where further edits to this guidance will surface first.