Mary Landesman, a senior security researcher at San Francisco-based ScanSafe, said it's more likely that the massive lists -- which include approximately 30,000 credentials from Hotmail, Gmail, Yahoo Mail and other sources -- were harvested by botnets that infected PCs with keylogging or data stealing Trojan horses.
Landesman based her speculation on an accidental find in August of a cache of usernames and passwords, including those from Windows Live ID, the umbrella log-on service that Microsoft offers users to access Hotmail, Messenger and a slew of other online services.
That cache contained about 5,000 Windows Live ID username/password combinations, said Landesman, who found the trove while researching a new piece of malware. "From the organisation [of that cache] and what the data looked like in raw form, I think it's more likely that this latest was the result of keylogging or data theft, not phishing," Landesman said.
She dismissed the idea that the passwords had been collected in a large-scale, industry-wide phishing attack, as Microsoft and Google both maintained.
"Another indicator is the sheer number of compromised accounts," Landesman said, referring to the two lists that have gone public. "Phishing is not generally a wildly successful scam, it doesn't have a big return. People are more savvy about phishing than we give them credit for."
Instead, it's more logical to assume that the passwords were acquired by botnet operators, who hijack PCs using security exploits, then later plant data-stealing malware on those machines. "That's a much more realistic source," said Landesman.
"Regardless [of] what the final intent is of a botnet, one of the core capabilities of every botnet is the harvesting of email credentials. If it looks like a horse, it's a horse, it's not a zebra."
Landesman's theory contradicts not only Microsoft and Google, but also the Anti-Phishing Working Group (APWG), an industry association dedicated to fighting online identity theft. On Monday, the APWG's chairman, Dave Jevans said a phishing attack that garnered thousands of passwords was do-able. "It's not outside the realm of possibility," he said then.
Also against the phishing explanation, argued Landesman, is the fact that the second list -- approximately 20,000 passwords -- contained usernames from not just Hotmail, but also Gmail, Yahoo Mail, Comcast, EarthLink and others. "That makes [the purported phishing campaign] a much broader attack across multiple services."
Her first thought when she read about the compromised Hotmail accounts was of the cache of credentials she'd found two months before. "Those public lists reminded me of the lists I found," she said. "It was definitely not a complete list, but seemed to be an advertisement for what this [hacker] had to offer."
The hacker was either inexperienced, or none too bright: The data was not password-protected, which is the norm for credential caches.
Landesman's theory is not just an academic exercise, she maintained.
"Everyone who suspects that their account has been compromised should change their password," she said, repeating advice by Microsoft, Google and other security experts. "But if, after changing their password, they have another reoccurrence where they see their account being used to email spam, or they again can't access their account, then they need to suspect that there's a local infection on their PC."