Criminals often register their own domain name to perform phishing attacks. Unlike the other common phishing site scenarios (including hacked servers, open redirects, and abuse of free webhosting), phishing sites that have their own domain name can be harder to remove, because the website owner and domain owner is the fraudster. Only the hosting and DNS providers and the domain registrar are able to take the site down and also likely to cooperate.
The operation of top-level domains is generally split between a registry, which operates the infrastructure that answers DNS queries, and registrars, which sell domain names and provide the process for owners to maintain their records. Registries generally are not directly involved in removing phishing domains, and refer those to the registrar through which the domain was registered.
However, it is relatively easy to become a registrar, so large numbers of hosting companies, web design firms and domain name resellers are able to handle registrations. Registrars may not all respond quickly to abuse complaints. And in unusual cases registrars themselves may be involved in illegal activity.
There is a particular problem with so-called fast flux phishing attacks. Here the attacker uses a large pool of compromised hosts — often personal computers on DSL connections — and from these randomly chooses a number to act as web servers to host the phish (and also some to act as DNS servers for the phishing domain). The set of hosts used to support the phishing site is changed regularly, so efforts to contact the owner of one hacked system would at best cause the phishing site to be temporarily unavailable. ICANN (which hands out the contracts to operate generic top level domains including .com) published a report earlier this year looking at whether it should intervene to encourage adoption of more effective policies by registrars to prevent the abuse of fast-flux setups; but it seems reluctant to compel registrars to stop a practice that may also have some legitimate uses.
The one common point for any phishing attack is the URL sent to victims. In the case of fast-flux attacks, the owner of the domain will not cooperate and there are too many hacked systems hosting the phish for contacting the hosting provider to be effective. The only place where the attack can be quickly stopped is for the registrar or registry to suspend its domain name.
The policies of the DNS registry for the top-level-domain containing the site are therefore important. The most practical indication of the relative success of these policies is to look and see which top-level-domains (TLDs) are most often used for whole-domain phishing attacks:
The high placement of .tk is unsurprising, given that it is possible to register .tk domains for free that redirect to any URL, completely anonymously. .com is the most common TLD for phishing domains, perhaps due to the ease of registering .com domains, and because the large number of registrars for .com domains gives an opportunity for fraudsters to look for registrars with weak checks or that respond slowly to abuse reports.
Finding an efficient escalation process in the case where the registrar is slow to cooperate will be the key to reducing the number of domains registered for phishing. The system that was designed to deal with domain disputes around ownership and trademarks is now looking too cumbersome when dealing with the problem of phishing attacks, where fast responses are essential to minimising fraud.