201,165 domains were born on Friday, September 4, 2026
On Friday, September 4, 2026 the census compared 1,052 zone files holding 75,861,702 domains against the day before. 201,165 names were new (0.265% of the universe), 156,978 were gone, net +44,187. Every new name was probed once over HTTP, honouring robots.txt, to see what was behind it. See the newest day.
New domains per day
What was behind the new names
- Active website31,292 15.6%
- Parked or for sale3,481 1.7%
- Registered, no site yet22,575 11.2%
- Nothing there at all57,088 28.4%
- Couldn't tell86,729 43.1%
What the 31,292 live sites are about
Where the new names landed
| TLD | Born | Gone | Flagged /1k |
|---|---|---|---|
| .xyz | 69,087 | 53,239 | 4.2 |
| .top | 21,024 | 22,137 | 2.7 |
| .online | 10,098 | 10,412 | 4.1 |
| .shop | 10,088 | 12,688 | 1.1 |
| .org | 10,074 | 6,923 | 2.3 |
| .site | 7,192 | 4,485 | 2.5 |
| .store | 5,527 | 4,875 | 3.3 |
| .info | 5,359 | 7,665 | 2.6 |
| .cyou | 4,606 | 554 | 1.7 |
| .app | 4,318 | 1,098 | 3.5 |
| .vip | 3,821 | 2,502 | 1.3 |
| .dev | 3,106 | 871 | 1.9 |
| .asia | 2,608 | 646 | 0.0 |
| .click | 2,200 | 1,126 | 10.0 |
| .pro | 2,191 | 1,754 | 2.3 |
| .space | 1,924 | 1,050 | 3.6 |
| .lol | 1,871 | 396 | 8.6 |
| .icu | 1,847 | 1,114 | 13.0 |
| .biz | 1,839 | 657 | 2.2 |
| .live | 1,568 | 1,167 | 7.7 |
| .sbs | 1,498 | 1,289 | 4.7 |
| .tech | 1,135 | 626 | 5.3 |
| .cloud | 1,046 | 583 | 2.9 |
| .fun | 992 | 761 | 13.1 |
| .cfd | 959 | 912 | 12.5 |
Top 25 of 525 TLDs with a birth or death that day. Full table: /api/zone/census/tlds?day=2026-09-04.
Who is parking them
What the live sites are built with
Writing system of the live sites
How old the departed names were
Risk signals
Every one of the 377,822 births was scored on its name, its nameservers and registrar, and the page it served. 12,462 scored at all; 47 critical, 1,080 high, 1,779 medium, 9,556 low. The top 300 were screenshotted and shown to a model, which called 145 of them phishing, scam or abuse and 91 benign.
TLDs with the most flagged births per 1,000
Signals that fired most
Brands most often echoed in new names
Registrars of flagged names
Nameserver providers of flagged names
Worth a second look
Signals, not verdicts. A domain appears here because heuristics matched its name, its infrastructure or the page it served on its first day. Where a model has looked at it, that is shown as an opinion with a confidence, not as a finding. Nothing on this page says a site is malicious. Read the evidence and decide for yourself.
Own one of these names and think the listing is wrong? Tell us and a person will look.
Model opinion: looks like abuse (target: Adobe), confidence 76%. An opinion, not a finding.
Model opinion: looks like phishing (target: PayPal), confidence 98%. An opinion, not a finding.
Model opinion: looks benign, confidence 60%. An opinion, not a finding.
Model opinion: looks like phishing (target: Coinbase), confidence 93%. An opinion, not a finding.
Model opinion: looks like a scam (target: OpenAI (ChatGPT)), confidence 92%. An opinion, not a finding.
Model opinion: looks benign, confidence 68%. An opinion, not a finding.
Model opinion: looks like abuse (target: Shopee), confidence 58%. An opinion, not a finding.
Model opinion: looks like abuse, confidence 82%. An opinion, not a finding.
Model opinion: looks like phishing (target: Google), confidence 62%. An opinion, not a finding.
Model opinion: looks benign, confidence 85%. An opinion, not a finding.
Model opinion: suspicious, confidence 55%. An opinion, not a finding.
Model opinion: looks like a scam (target: Roblox), confidence 76%. An opinion, not a finding.
Model opinion: looks like a scam, confidence 87%. An opinion, not a finding.
Model opinion: looks benign, confidence 92%. An opinion, not a finding.
Model opinion: looks benign, confidence 87%. An opinion, not a finding.
About this data
The census diffs the zone files ICANN's Centralized Zone Data Service publishes for about a thousand gTLDs, plus a sample of .com and .net, against the previous day. A birth is a name present today and absent yesterday; a death is the reverse. Births are probed once over HTTP with an identified user agent, robots.txt honoured, and classified by what came back. Subjects are assigned by a language model reading the page text. The risk scorer is deterministic and additive; its weights and word lists are public in the API's reasons field.
Aggregates and named rows are free as JSON: API docs. Per-domain answers are also on every domain report. Names in .com and .net are counted in every aggregate but not listed by name until the redistribution terms of that zone's access agreement are confirmed.