Google Just Acknowledged Vibe-Coded Websites – Here’s What It Means for SEO

Something shifted this week. Google’s John Mueller dedicated a chunk of his latest Search News video to what he called “vibe-coded websites”, sites built primarily with AI tools like Bolt, Lovable, Cursor, or v0.

His verdict? They’re essentially normal websites and generally fine for search.

That’s a meaningful moment. It’s Google officially acknowledging that AI-built sites are now mainstream enough to address directly, and giving them a degree of legitimacy in the process.

But Mueller didn’t just give them a clean bill of health and move on. He flagged several things worth paying attention to. And having spent time auditing AI-built sites myself recently, his list maps closely to what I keep finding in the wild.

Here’s what you need to know.

What Are Vibe-Coded Websites?

The term “vibe coding” was coined by AI researcher Andrej Karpathy earlier this year. It describes a way of building software, including websites, where you describe what you want in plain language and let an AI tool do the actual coding.

Tools like Bolt, Lovable, v0, and Cursor have made this genuinely accessible. You don’t need to know HTML, CSS, or JavaScript. You describe your business, your goals, the look and feel you’re after, and the tool builds something that often looks and functions like a proper website.

For small businesses, freelancers, and anyone who previously couldn’t afford a developer, this is a big deal. The barrier to having a decent web presence has dropped dramatically.

But a site that looks good in a browser and a site that performs well in search are not always the same thing.

What Mueller Actually Said

In his Search News video, Mueller made a few specific points about vibe-coded sites that I think deserve attention beyond the headline.

The first was about content. He said the most important thing is making sure what you’re publishing actually adds value to the web. “It’s easy to just create a site nowadays,” he said, “but if you want people to appreciate it, make sure it’s actually good.”

That’s not a new message from Google. But it’s worth hearing in this context. AI tools make it trivially easy to fill pages with words. That’s not the same as filling pages with something useful.

The second point was technical. He specifically called out canonical tags, noting that sites should include the full URL of each page including the domain name. This sounds minor but it’s a real issue, many AI site builders generate canonical tags that reference a path rather than an absolute URL, which creates ambiguity Google sometimes resolves in ways you wouldn’t want.

Third, he flagged JavaScript rendering. If your site is built on a framework like React or Next.js, which several AI builders use under the hood, Google may see something very different from what you see in a browser. He recommended testing pages using Google’s own tools to confirm they’re rendering properly.

Finally, he recommended adding the site to Search Console from day one.

What I See When I Audit AI-Built Sites

Mueller’s list is accurate in my experience. Here’s what I’d add from auditing these sites myself.

Content quality is the main problem. The technical bones of AI-built sites are often surprisingly solid. A well-prompted build handles the basics. What tends to fall short is the content. Generic descriptions, no real depth, nothing that demonstrates genuine expertise or experience. Google has been leaning harder on E-E-A-T signals for a couple of years now and thin AI content doesn’t satisfy those signals.

Structured data needs a human check. Some AI builders do generate structured data automatically and sometimes it’s fine. But I’ve seen incorrect schema types, missing required fields, and structured data that validates but describes the page inaccurately. Always run your pages through Google’s Rich Results Test before you call a site done.

Internal linking is an afterthought. Most AI-built sites have weak internal link structures. Pages exist but don’t connect to each other in a way that helps search engines understand the site’s structure or distributes authority. This is easy to fix but easy to miss if you haven’t thought about it.

Page titles and meta descriptions are often generic. AI tools usually generate these but they tend to be bland. A title like “Home, Business Name” tells Google and potential visitors very little. Take the time to write titles that reflect what each page is actually about and include the terms people would search for.

The SEO Checklist for AI-Built Websites

If you’re launching a site built with AI tools, or auditing one for a client, here’s what I’d check before going live.

Technical foundations

• Canonical tags use absolute URLs including the domain (https://yourdomain.com/page/ not /page/)
• If built on React or Next.js, pages render correctly for Google, check with URL Inspection in Search Console
• XML sitemap exists and has been submitted to Search Console
• robots.txt is present and not accidentally blocking important pages
• Page speed is acceptable, run through PageSpeed Insights

Content

• Every page has a unique, descriptive title tag (not just the business name)
• Meta descriptions are written and relevant to each page
• H1 heading is present on every page and reflects the page topic
• Content goes beyond the generic, it should say something specific and useful
• No duplicate content across pages

Schema and structured data

• Structured data is present where relevant (LocalBusiness, FAQPage, etc.)
• Validate with Google’s Rich Results Test
• No errors or missing required fields

Search Console

• Site verified in Google Search Console
• Sitemap submitted
• URL Inspection run on key pages to confirm indexing

Should Small Business Owners Be Using These Tools?

Mueller’s closing comment on this was interesting. He said vibe-coded websites might not take over the popular web just yet, but if you work with smaller clients it can be worth getting some practice with these tools.

I’d agree with that framing. For small businesses that need a credible web presence quickly and don’t have budget for a full build, AI tools have genuinely crossed a threshold. The output is good enough.

But “good enough to launch” and “optimised for search” are different things. The speed and ease of AI site building can create a false sense that the job is done when actually the SEO work is just starting.

The businesses I see getting results from AI-built sites are the ones that treated the AI build as the starting point, not the finished product. They went back and improved the content, set up tracking, fixed the technical issues, and gave Google something worth ranking.

The ones that struggled launched, did nothing else, and wondered why nothing happened.

Frequently Asked Questions

Are vibe-coded websites bad for SEO?

Not inherently. Google’s John Mueller confirmed in April 2026 that AI-built websites are essentially normal websites and can perform fine in search. The SEO issues that tend to appear on these sites, thin content, missing canonicals, JavaScript rendering problems, are fixable and not unique to AI-built sites.

What is vibe coding?

Vibe coding refers to building websites or software by describing what you want in plain language and letting an AI tool generate the code. Tools like Bolt, Lovable, Cursor, and v0 have made this accessible to people with no coding background.

Do AI-built websites rank on Google?

Yes, they can. The fundamentals of ranking haven’t changed, Google looks for pages that are well-structured, load quickly, cover a topic with genuine depth, and earn links and trust over time. An AI-built site that gets those things right can rank. One that doesn’t will struggle, regardless of how it was built.

What should I check before launching an AI-built website?

The key checks are: canonical tags using absolute URLs, Search Console setup and sitemap submission, JavaScript rendering tested if using React or Next.js, structured data validated, unique title tags on every page, and content that goes beyond the generic.

Does Google treat AI-generated content differently?

Google’s stated position is that it evaluates content on quality, not how it was produced. AI-generated content that is accurate, useful, and demonstrates expertise can perform well. AI-generated content that is thin, generic, or inaccurate will not, regardless of whether a human or a machine wrote it.