Chris Kelly, chief privacy officer at Facebook, told delegates at the annual eCrime Congress in London that the firm has a range of privacy and security-enhancing tools in place.
The first line of defence, he explained, is a system of automatic scanning tools which check for fake accounts which might be sending spam or malicious content. Key signs include accounts which reject too many 'friend' requests, or carry out an unusually large number of searches.
"Our innovative privacy controls include behind-the-scenes scanning for fake accounts," said Kelly. "We take down thousands of accounts a week, and on the rare occasions where they get through, we can build protection into our system [for the next time]."
Kelly also spoke of "the cops on the beat", a user operations and investigations team which complements the automated systems by deploying clean-up tools on compromised or fake accounts.
The team handles hundreds of thousands of user contacts a week. "The challenges are many but our tools are many also," he added.
However, statistics published last October by messaging security vendor Cloudmark suggested that 20 to 40 per cent of new profiles created on Facebook could be fake.
Neil Cook, European head of technology services at Cloudmark, said that there could be many more fake profiles because Facebook does not check for "sleeper" profiles created by spammers at the same time as they register active accounts.
"By the time firms have shut one down, the spammer has moved on to another profile they have as a back-up, which means the [social network] is constantly fire-fighting," he explained. "It would be interesting to know what Facebook does about sleeper profiles."
Facebook has come in for heavy criticism in recent weeks over a series of privacy and security related matters, including the exploiting of Facebook accounts by hackers, and controversy surrounding new terms and conditions.