"Imagine that a specific web shop gets redirected to its identical rogue clone where users enter credit card information, delivery addresses and phone numbers for couriers' convenience," Botezatu said. "In 20 minutes, the attacker can easily collect a huge pool of data, then change the IP addresses back to the original ones. In some cases, website owners may actually not spot the problem at all."
In another scenario, an attacker could change the MX (mail exchanger) record for the domain name and route all email communication to a mail server under his control, Botezatu said.
The credit card and billing information provided by customers with paid accounts is not displayed on the website, so it was not at risk of being compromised as a result of this attack, Prince said.
"These type of phishing attacks have been common among registrars and web hosts for some time," Prince said. "It is only because of CloudFlare's increasing scale, now with more than half a million customers, that we've become an interesting target."
The company was able to determine that only around 200 customers had clicked on the rogue link in the spoofed emails before the phishing website was taken down. All of those customers had their passwords reset, Prince said.
For the past few months the company has been working on implementing a two-factor authentication feature that would prevent account abuse even when the account password is compromised, Prince said. "We plan to turn it on in the next few days. Likely tomorrow."
CloudFlare had already implemented the Sender Policy Framework (SPF) and DomainKeys Identified Mail (DKIM) technologies to allow email providers to verify the sender's identity and block emails that spoof CloudFlare as their source, Prince said. "We will be adding a DMARC [Domain-based Message Authentication, Reporting and Conformance] policy to further enhance this protection."