Skip to content
Security & Infrastructure4 min read

New gTLDs in 2025: 1 in 5 Registrations Was a Scam

A June 2026 Interisle report found that up to 20% of the ~85 million new gTLD domains registered in 2025 were malicious — and enforcement barely touches them. Here's what that means if you're building fintech or Web3 products.

By TRAGenX Desk

Share

DNS is supposed to be boring, load-bearing infrastructure — you register a name, it resolves, done. But a new report puts a number on something builders have felt anecdotally for years: a meaningful share of new domain registrations exist purely to run scams, and almost nobody is taking them down.

The numbers

  • Interisle Consulting Group's report, Malicious Registrations in the Domain Name Market, published June 1, 2026, found that of the roughly 85 million new generic top-level domain (gTLD) registrations made in 2025, 8.5 million (10%) had already appeared on security blocklists by mid-May 2026.
  • Interisle's own projections — accounting for undetected and not-yet-blocklisted registrations — push the real abuse rate toward 20%, or an estimated 16.8 million domains.
  • Abuse is heavily concentrated: 13 TLDs exceeded 50% blocklist rates, led by .LOCKER (72.9%), .LGBT (72.2%), and .TOWN (70.2%), and five registrars were responsible for half of all blocklisted domains.
  • Enforcement lags badly — only 7.4% to 16.3% of blocklisted domains were ever suspended.

Terence Eden put it bluntly, in a post Simon Willison flagged after seeing Andrew Campling's analysis on RIPE Labs: "One in five newly registered domains with a gTLD are scams."

Why this matters if you're building fintech or Web3 products

Trading platforms, wallets, and exchanges are exactly the kind of high-value, high-urgency brands that phishing infrastructure loves to impersonate — a lookalike domain with a fresh SSL cert is cheap to stand up and short-lived by design, which is precisely why blocklist-based takedown struggles to keep pace. The Interisle numbers quantify the base rate of that risk across the entire gTLD namespace, not just your brand's corner of it.

There's a second, more novel angle for anyone shipping AI-assisted or agentic systems: an LLM agent that autonomously fetches a URL, follows a link in a support ticket, or pulls off-chain data for an on-chain process has no innate skepticism about domain age or registrar reputation. It resolves and reads. If a fifth of new registrations in the namespace are adversarial, treating a resolvable domain as trustworthy input is a mistake — the same trust-boundary discipline you'd apply to any other untrusted data belongs on URLs an agent visits unsupervised.

What it changes for how you ship

  • Treat TLD and registrar as a fraud-detection feature, not just a branding decision — "newly registered on a historically high-abuse TLD" is a legitimate, cheap signal for a KYC/AML or account-fraud pipeline.
  • If your product includes agentic browsing, auto-fetching, or link-following (support bots, research agents, oracles), don't let resolution alone confer trust — pair it with domain-age and blocklist checks the way you'd validate any other external input.
  • Monitor for lookalike registrations of your own brand rather than assuming users will notice a wrong TLD or an extra hyphen — by the time a scam domain hits a blocklist, most of its damage is already done.

None of this means DNS is broken beyond repair — it means the economics currently favor high-volume, low-friction registration over verification, and builders who treat that as a known, measurable risk will catch more fraud than those who don't.

FAQ

Frequently asked questions

What counts as "DNS abuse" in this report?
Interisle measured domains that appeared on security blocklists shortly after registration — typically phishing, malware, or scam sites — as a proxy for malicious intent behind the registration itself.
Which report is this based on, and when was it published?
Interisle Consulting Group's "Malicious Registrations in the Domain Name Market," published June 1, 2026, analyzing 2025 gTLD registrations against blocklist data as of mid-May 2026.
Should crypto and trading platforms do anything differently because of this?
The report doesn't break down abuse by target industry, but phishing infrastructure broadly targets high-value brands like exchanges and wallets — so brand-monitoring for lookalike domains and treating domain reputation as a fraud signal are reasonable, low-cost defenses regardless of what any single report says about your sector specifically.

Sources

Share

Read next