Google doesn't care that you used AI. Google cares that your content is rubbish.

Let's kill the biggest myth first. Google has not said "AI content will be penalised." What Google has said, repeatedly, is that low-quality content will be penalised regardless of how it was made. The March 2024 core update wiped out a lot of AI-generated content, but it also wiped out a lot of human-generated content that was equally thin and unhelpful.

The pattern Google penalised was specific: sites that used AI to mass-produce hundreds of shallow articles targeting long-tail keywords, with no original insight, no expertise, and no reason to exist beyond capturing search traffic. If that sounds like your strategy, stop. If you're using AI to produce and improve genuinely useful content, you're fine.

Now, here's where AI actually helps with SEO, and where it'll get you in trouble.

Where AI genuinely helps

Keyword research and clustering

This is where AI saves the most time with the least risk. semrush and ahrefs both have AI-powered keyword clustering tools that group related keywords by search intent. Instead of staring at a spreadsheet of 500 keywords trying to figure out which ones belong on the same page, the AI does it in seconds.

The workflow I use: Export keyword data from Semrush or Ahrefs for your topic Paste the keyword list into claude with the prompt: "Group these keywords by search intent. For each group, identify: the primary keyword, the intent (informational, transactional, navigational), and suggest a content format (blog post, landing page, comparison page, how-to guide)." Use the output to plan your content calendar

This takes about 10 minutes and replaces what used to be a full afternoon of manual categorisation. The AI is surprisingly good at identifying intent nuances, like distinguishing between "best CRM software" (comparison intent) and "CRM software features" (informational intent).

Content gap analysis

Another strong use case. Feed your existing content list and your competitors' content into Claude or ChatGPT and ask it to identify topics your competitors cover that you don't. This replaces the manual process of crawling competitor sites and cross-referencing topics.

Prompt that works: "Here are the blog post titles from my site: [list]. Here are the blog post titles from my top 3 competitors: [lists]. Identify topics that at least 2 competitors cover but I don't. For each gap, suggest a specific article title and outline."

Meta description generation

Writing meta descriptions is tedious. The AI is good at it. Feed it the page content and ask for a meta description under 155 characters that includes the target keyword and a reason to click. I've tested this against human-written meta descriptions and the click-through rates are comparable.

surfer-seo automates this within their platform, scanning your content and generating optimised meta descriptions that match the competitive set. It's not magic, but it saves genuine time when you have 50 pages that need updated meta descriptions.

Schema markup

This is an underrated use for AI. Structured data (JSON-LD schema markup) is fiddly to write by hand and easy to get wrong. Give Claude your page content and ask it to generate the appropriate schema markup. It handles FAQ schema, HowTo schema, Article schema, and Product schema reliably.

Prompt: "Generate JSON-LD structured data for this page. The page is a how-to guide about [topic]. Include HowTo schema with these steps: [steps]. Also include an FAQ section with these questions: [questions]. Output valid JSON-LD that I can paste into the head of my HTML."

I've been doing this for six months and haven't had a single schema validation error from Google's testing tool. The AI is just good at structured formats.

Content outlining

Getting an AI to outline an article based on what's currently ranking is genuinely useful. The key is using it as a starting point, not a final product.

My workflow: Search the target keyword in perplexity to see what's currently ranking and what those articles cover Ask Claude to create an outline that covers the same subtopics but from a different angle Manually add my own opinions, experience, and original insights to the outline Write the article using the outline as a skeleton

This produces content that's competitive with what's ranking but brings something new to the table. Google's helpful content system specifically looks for original insight, and this workflow ensures you have some.

Where AI will hurt your rankings

Publishing raw AI content

Don't. Just don't. I've tracked dozens of sites that published bulk AI content in 2025, and the pattern is consistent: initial traffic bump as Google indexes the pages, followed by a steady decline over 3-6 months, sometimes ending in a manual action or significant ranking drop.

The problem isn't that Google can "detect" AI content (they probably can, but that's not the issue). The problem is that raw AI content is, by definition, derivative. It's a statistical remix of everything else on the internet. It adds nothing new. Google's entire ranking philosophy now is about surfacing content that provides genuine value beyond what already exists.

If you use AI for a first draft, you need to edit it substantially. Add personal experience. Add original data. Add opinions. Add anything that makes it more than a reworded version of the current top 10 results.

Automated link building outreach

AI-generated outreach emails are getting auto-deleted by everyone. We all recognise them now. The generic praise ("I really enjoyed your recent article about..."), the formulaic pitch, the impersonal sign-off. If your link building strategy is "have AI send 500 emails a day," you're burning bridges with people you might actually want relationships with.

Ignoring E-E-A-T signals

Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) puts a premium on content created by people with genuine experience. An AI doesn't have experience. It can write about hiking but it has never hiked. It can write about cooking but it has never tasted anything.

This matters for SEO because Google increasingly rewards content with clear author attribution, demonstrated expertise, and first-hand experience. If your content reads like it was written by someone who has never actually done the thing they're writing about, Google will rank it below content from someone who has.

The fix: Use AI for research, structure, and drafting. But the opinions, anecdotes, and expertise need to come from you.

The AI Overviews elephant in the room

Google's AI Overviews (the AI-generated summaries at the top of search results) now appear on roughly 30-45% of informational queries. This changes the SEO calculus significantly.

The impact: For queries where Google shows an AI Overview, the click-through rate to organic results drops by 30-60%, depending on the query type. If Google's AI can fully answer the question from the Overview, many users never click through to any website.

What this means for your strategy:

Informational, easily-answered queries ("what is CRM software," "how does DNS work") are becoming less valuable as traffic sources. The AI Overview answers them and users move on.

Queries that require depth, comparison, opinion, or experience ("best CRM for small agencies," "is Salesforce worth the price," "CRM implementation mistakes to avoid") still drive clicks because the AI Overview can't fully satisfy the user's intent.

The strategic shift: Target queries where the answer requires nuance, personal experience, or subjective evaluation. These are harder for AI Overviews to fully address, and users will still click through to read the full article.

ahrefs and semrush both now show whether a keyword triggers an AI Overview, which helps you prioritise keywords where organic clicks are still healthy.

The honest take

The best SEO strategy in 2026 is the same as it was in 2016, just faster with AI tools. Write genuinely useful content. Build real authority. Create stuff that provides value beyond what a Google search snippet (or AI Overview) can deliver.

AI makes the mechanical parts of SEO faster: keyword research, clustering, meta descriptions, schema markup, content outlining. Use it for those things without hesitation.

AI does not make the hard parts of SEO easier: having original insights, building genuine authority, earning links from people who actually value your content. Those still require being a real person with real expertise who puts in real effort.

The people who'll win at SEO in 2026 are the ones using AI to do the boring stuff faster so they can spend more time on the stuff that actually matters. And no, the irony that I used AI to help write this article about AI and SEO is not lost on me.