A .ai Domain Dispute Shows the Limit of Trademark Gravity

TL;DR: A trademark does not give a company automatic control over the matching .ai domain, and AI builders should treat naming as a legal, product, and operating risk from day one.

What actually happened with hyperfly.ai?

Primary source for this note is Domain Name Wire’s report, “James Booth successfully defends .ai domain in cybersquatting dispute.”

Domain Name Wire reported that Do or Die, Inc., doing business as Hyperfly, filed a World Intellectual Property Organization complaint against James Booth over hyperfly.ai. Hyperfly sells combat sports gear. Booth, described by Domain Name Wire as an AI domain investor, defended the domain successfully. A three-person WIPO panel found against the complainant, according to Domain Name Wire.

That is the whole surface story. But the useful part is not “domain investor wins” or “brand loses.” It is this: matching strings are not enough.

In cybersquatting disputes, a complainant usually needs more than a trademark and a domain that looks similar. The hard parts tend to be whether the registrant lacked legitimate interests and whether the domain was registered and used in bad faith. Those details matter because .ai names are weirdly overloaded now. A word can be a brand, a product idea, a generic-ish term, a speculative asset, or an AI company placeholder, sometimes all at once.

Hyperfly is a real brand in combat sports. Hyperfly.ai also sounds like a plausible AI product name. That collision is exactly where the .ai boom creates friction.

Why does this matter for AI builders?

Because naming has become part of the build stack.

A few years ago, a startup could pick a name, grab a decent .com variant, ship, and clean up the trademark work later. That was never ideal, but it was common. In AI, the namespace is tighter. Short .ai domains are treated like scarce inventory. Product teams want names that sound agentic, fast, technical, or magical. Domain investors got there early. Trademark owners may arrive later and assume the domain should be theirs.

This case is a reminder that those assumptions can fail.

It also cuts the other way. Owning a .ai domain does not mean you are safe to build under that name. A domain win at WIPO is not a trademark clearance. It does not mean a company can launch a product into a category where another party has strong rights. Domain disputes and trademark infringement lawsuits are different tools with different burdens, remedies, and costs.

For operators, the practical mistake is treating “available domain” as the same thing as “available brand.” It is not. The second mistake is treating “we have a trademark” as the same thing as “we can take that domain.” Also not true.

What is the useful lesson, not the domain drama?

The useful lesson is to separate three questions early.

Can you use the name without creating obvious legal exposure? Can you get a domain that customers will trust? Can you defend the name if the product works and copycats show up?

Those are related, but not identical. A founder may clear one and fail another. A model wrapper with a catchy name can get blocked by a trademark issue. A serious B2B tool can lose credibility because the clean domain is held by someone else. A domain owner can win a cybersquatting dispute but still have no product, no distribution, and no brand equity.

This is where I think the .ai conversation gets silly. People talk about domains as if they are destiny. They are not. They are routing, memory, trust, and sometimes status. Useful, but not magic. The product still has to earn the name.

For builders, do the boring work before the launch post. Run trademark searches. Check adjacent categories, not just your exact category. Look at the domain history. Save evidence of why you picked the name. If you buy a domain, document the transaction. If you are sitting on a name before launch, avoid parking pages that make you look like you targeted a specific brand. And if the name is core to the company, talk to a lawyer before customers, investors, and integrations make it expensive to change.

Practitioner’s Take: Pick names with a simple clearance checklist, not just taste. Before you ship, test three candidates for trademark risk, domain trust, and future defensibility. The catch most teams miss is timing: naming risk is cheap when it is a Notion doc and painful once it is in your GitHub org, auth flow, investor deck, and customer contracts.