Google's current guidance on generative AI says the quiet part plainly: using AI to make many pages without adding value for people may breach its scaled-content-abuse policy. The practical question for a premium brand is not whether an editor used AI. It is whether the finished page says something that could not have been published by a hundred other sites.
That is where commodity content becomes expensive. It can fill a calendar, but it makes a brand sound interchangeable at the exact moment a reader is deciding whether the brand has judgement worth trusting.
Google's guidance on generative AI content asks publishers to focus on accuracy, quality and relevance, including the title, meta description, structured data and image alt text. In May, Google also said its AI Search experiences were being adjusted to connect people with original content and direct links to useful sites. The direction is hard to miss: an answer without a distinct source behind it has less reason to exist. This analysis is supported by Google Search updates and Google generative-AI Search overview.
The bad unit of production is a page
A content team can still publish a technically correct page that adds nothing. It may repeat a familiar definition, reframe a press release, and place a keyword in a neat set of headings. The page looks complete. It is not useful.
For luxury technology, the risk is sharper because readers often arrive with a high-consideration question. They may be weighing privacy against convenience, trying to understand a feature before a purchase, or deciding whether a new AI workflow belongs near sensitive work. Generic reassurance does not answer any of those questions.
The better unit of production is a resolved reader decision. Before writing, an editor should be able to name it:
What does the reader need to decide or understand?
What evidence will change that reader's view?
Which facts belong to a primary source, rather than to brand prose?
What limitation or boundary would a careless article leave out?
If those questions have no answer, the page is probably a calendar entry rather than an editorial contribution.
Four tests before a draft enters production
1. Could this have been written for another brand?
Replace the brand name with a competitor's. If the article still works unchanged, it has not earned its place. A VERTU article does not need to turn into an advertisement, but it should offer a considered point of view about private technology, executive use, craftsmanship, or human service.
2. Is the source doing real work?
A link should support a claim, not decorate a paragraph. If an article says a platform introduced a new permission model, link to the platform's announcement. If it interprets that model, label the interpretation as an editorial judgement. This lets a reader distinguish what happened from what the writer thinks it means.
3. Does the article include the uncomfortable detail?
The useful detail is often a limit: an availability restriction, a permission requirement, a trade-off, or a task that should still be checked by a person. That is not a weakness in premium communication. It is how a reader recognises care.
4. Does the metadata tell the same truth as the body?
Google calls out metadata and image alt text alongside the article itself. A page should not promise a comparison, review, or product capability in the title and then deliver a broad essay. The description, heading, image, and body should describe the same piece of work.
A better role for AI in a premium editorial process
AI is useful for research organisation, question generation, first-pass structure and line editing. It is less useful as a substitute for a product source, a customer insight, or an editor willing to remove a weak idea.
That distinction matters for VERTU. Take Hermes Agent: the public product position is a private, context-aware assistant for selected devices, with user approval for significant actions and availability that varies by configuration and supported service. An article about personal agents should retain those boundaries. It should not turn a nuanced feature into a sweeping promise just because a fluent draft makes that easier.
The standard is modest. Publish fewer pages that carry a real point of view, show the source behind material claims, and leave the reader clearer than when they arrived. In an AI-heavy web, that is more distinctive than volume.
Frequently asked questions
Does Google ban content written with AI?
No. Google's guidance says generative AI can help with research and structure. Its concern is automated production that creates many pages without value for users.
What makes brand content non-commodity?
Specific evidence, a clear reader decision, accurate product boundaries, and an editorial point of view make a page harder to replace with generic copy.
Should every page mention the brand?
No. A brand mention should appear only where it gives the reader useful context. A forced product reference makes an informational article less credible.
