Spam email recently sent out on behalf of NTL pointed to a potentially serious vulnerability in the cable operator's online processing system.
Although NTL acted quickly to shore up the potential problem it serves as yet another example of why spam messages can damage an organisation's brand.
The offending message was sent to Matthew Garrett, of Cambridge University's Computer Laboratory, via a third party but pointing to a seemingly legitimate NTL offer.
"I wasn't terribly amused by this, but checked the Web site to see whether it did seem legitimate," Garrett explains. "While there, I noticed that there seemed to be a large amount of English embedded in the URL, which also appeared in the text of the page."
Garrett discovered that extra HTML tags could be inserted into the URL, and worse, this extra information was passed through the link to a supposedly secure ordering page.
"With the aid of Steven Murdoch, a member of Ross Anderson's security group here in Cambridge, we constructed a cross-site scripting exploit that could be embedded in the original url," Garrett explained.
"Normally this sort of attack can be used to obtain user's cookies, but since a page taking credit card numbers was involved this time it also allows for a hostile user to cause the credit card numbers (along with all the other personal information) to be sent to a site somewhere else."
But would people visit this URL?
"NTL have already demonstrated that at least one of their advertising contractors is happy to spam people, so it'd look just as legitimate," Garrett argues.
Just as well that NTL has now fixed these security vulnerabilities. Let's hope it stops paying third parties to spam people too.
NTL told Garrett that it had asked a third party to check that the addresses were opt-in, but that about 50 per cent of the addresses they were told were clean were in fact trawled from Usenet and the like.
NTL scripting errors serve as an example for other organisations. So what lessons can we learn?
The real issue, according to Garrett, is "taking information from a URL and embedding it in a page - it's pretty much impossible to do this without letting people insert HTML tags that let them embed scripts from other websites (for instance) which subvert the page functionality. If the page is collecting credit card numbers, then the consequences are fairly obvious."
"Website authors need to be more paranoid," he argues.
"There are people who are sufficiently competent to work out how to pass variables to you scripts, even if you don't think that it's likely. Doing this with a site that accepts credit cards is a desperately bad idea, but it can be used for attacks in any case.
"There should be no excuse for making this sort of mistake."
The use of spam to tempt people to a poorly constructed Web site only makes matters worse.
As Garrett notes: "Spamming sufficiently competent people is likely to piss them off enough to subvert your website." ®