Can Google tell if a blog is written by AI?

In February 2023, Google published official guidance clarifying that AI-generated content is allowed when it is helpful and demonstrates real expertise. In March 2024, Google formalized a 'scaled content abuse' spam policy, making clear that the problem is content created to manipulate rankings, not the tool that produced it.

What Google actually detects is the signals of lazy AI output, not the AI itself:

• -Generic phrasing that appears across thousands of competing articles.
• -No original data, examples, or first-hand experience (a core EEAT signal).
• -Hallucinated citations or factual errors.
• -Pages that summarize the SERP back to itself with no new insight.
• -Sites publishing dozens of AI drafts per day with no editorial layer.

AI is fine for drafting. The post still needs a human pass that adds something the SERP does not already have. That is the line between ranked and invisible.