The question comes up in almost every client conversation now: "Can we just use ChatGPT for the blog?" It's a fair question. AI writing tools are fast, cheap, and capable of producing grammatically correct, well-structured text at scale. But the honest answer is more nuanced than yes or no — and getting it wrong in either direction has real consequences for your rankings.

Google's official position, confirmed multiple times through 2024 and into 2025, is that it does not categorically penalise AI-generated content. What it penalises is content that is unhelpful, low-quality, and produced primarily to manipulate rankings rather than serve searchers. The production method is not the variable. The usefulness is.

What did Google's 2024 updates actually do to AI content?

The March 2024 core update was the clearest signal yet. Google confirmed it targeted "unhelpful, unoriginal content at scale" — which in practice meant sites that had flooded the web with AI-generated articles designed to capture search traffic rather than answer real questions. Thousands of sites lost 50-90% of their organic traffic in a matter of days. The common thread was not AI authorship — it was mass production of content that lacked genuine depth, specific expertise, and real-world insight.

Where AI content actually helps SEO

Used strategically, AI is genuinely useful for several parts of content production that don't require human expertise.

AI is a production tool, not an expertise substitute. The content that ranks in 2025 demonstrates things an AI can't generate: original data, first-hand experience, and specific answers to questions nobody else has answered.

Where AI content actively hurts SEO

The danger zone is using AI as the primary (or only) author at scale without expert review. Generic AI output tends to be accurate but vague — it covers the topic without saying anything specific. Google's E-E-A-T quality standards specifically reward the "Experience" component: content that demonstrates first-hand knowledge, real client examples, original data, and specific context a generalist cannot provide. A 1,500-word AI article on "how much does SEO cost" that never quotes real figures, never references actual client outcomes, and never acknowledges market variation is the textbook example of content that passes a spelling check and fails a quality review.

The second danger zone is duplicate perspective. When everyone uses the same AI tools and prompts, the output converges toward the same angle. AI has been trained on existing web content, so it tends to reproduce the consensus view. If your content says exactly what every other article on the topic says, in the same structure, with the same examples, there's no reason for Google to rank yours above the established ones.

What the AI Overview era means for content quality

With Google's AI Overviews now surfacing direct answers above organic results for many queries, the incentive for content that earns citations has shifted dramatically. To be cited in an AI Overview, your content needs to be specific, structured, and demonstrably authoritative — not just keyword-present. This is actually good news for businesses that invest in real expertise: AI-generated generic content has no path to AI Overview citations. Original insight and specific, structured answers do. Understanding what kind of content you need for good SEO is the complement to this — the question is not just whether AI wrote it, but whether it meets the standard regardless.

A practical decision framework

Before publishing any piece of AI-assisted content, ask three questions. One: does this contain specific information a generalist AI wouldn't know — real client examples, original data, specific numbers, first-hand observations? Two: does it answer the searcher's question directly and completely, or does it feel like it's building up to an answer without delivering it? Three: if you removed the AI's contribution and kept only the human additions, would there still be something uniquely valuable left? If the answer to any of these is no, the content is not ready. This is also how tracking your SEO progress becomes useful — if AI-heavy pages are getting clicks but no engagement, the content is failing the human test even if it's passing Google's crawler.

Will Google ever be able to detect AI content?

Technically, yes — AI watermarking and detection tools exist and are improving. But Google's focus has consistently been on output quality, not production method. The practical risk is not getting "caught" for using AI; it's publishing content that fails quality signals over time and accumulates thin-content penalties. Detection is a secondary concern. Quality is the primary one.

Should I disclose that content is AI-assisted?

Google does not require disclosure. For most business content — blog posts, service page copy, FAQs — disclosure is neither expected nor particularly relevant to the reader. For content that depends heavily on personal experience or professional expertise (medical advice, legal guidance, financial planning), transparency about authorship matters more, and human expert review is non-negotiable regardless of drafting method.

What's the right ratio of AI to human in content production?

There's no universal rule, but a useful heuristic is: the higher the E-E-A-T requirement for a topic, the more human expert involvement the content needs. A blog post explaining what a redirect chain is can be largely AI-drafted and human-polished. A piece comparing SEO strategies for different business models, citing specific client results, needs a human practitioner's voice at the core. Let expertise requirements drive the ratio, not convenience.

Keep reading

AI content strategy is one piece of the broader content picture. What kind of content you need for good SEO covers the types, depth, and structure that actually earn rankings. And how long does SEO take is worth reading alongside this — because the compounding nature of SEO means content quality decisions made today show up in traffic data six months from now.