Twitter has added a new service that detects malicious URLs in an effort to quell the rise in spam and phishing on the microblogging social network.

The new security feature ultimately will scan all URLs before they hit the Twitter feed, but initially is only doing so for URLs sent via Twitter direct messages [DMs] and email notifications about DMs. Twitter is using its own URL shortener for these links: "For the most part, you will not notice this feature because it works behind the scenes but you may notice links shortened to twt.tl in Direct Messages and email notifications," said Del Harvey, Twitter's director of trust and safety, in a blog post last night.

Twitter's security feature comes amid new data revealing the level of abuse on the social network: One in eight Twitter accounts last year was malicious, suspicious, or suspended, according to a report issued today by Barracuda Networks. The surge in celebrities joining Twitter in 2009 resulted in a major jump in spam, phishing, and other abuse on the site, according to the report.

And those numbers have remained steady to date. "We are still seeing Twitter identify 3 to 4 percent of Twitter accounts as malicious. And, meanwhile, 9 to 10 percent of accounts on Twitter are actively engaging in malicious activity," says Paul Judge, chief research officer at Barracuda.

Twitter's abuse rate increased 66 percent during what Barracuda calls the "Twitter Red Carpet Era," the period during November 2008 to April 2009 when a wave of celebrities joined the social network. A copy of the full report is available here (PDF).

Chet Wisniewski, senior security adviser at Sophos, says Twitter's move to defend against attacks on its users is good news, although the announcement doesn't provide much detail on how the service works. "I did some testing, and it appears they are only converting links to their shortener, http://twt.tl, for email notifications of direct messages at this time," Wisniewski says. "Hopefully we will see Twitter partner with more security organizations to help stop spam, viruses, and other scams as well."

As of this posting, Twitter had not responded to a request for an interview.

Barracuda's Judge says protecting Twitter DMs is a good start, but that's not where the majority of malicious links are conducted on Twitter. "We've seen the majority of activity in the public time line, with attackers trying to take advantage of popular trending topics," Judge says. "At least Twitter is acknowledging the [malicious URL] problem."

Judge says it's unclear why Twitter needs a URL shortener to safeguard URLs. "I almost wonder if they wanted to have a URL shortener and are using security as a reason [to launch it]," he says.

Twitter should be conducting more analysis of links being distributed around its platform, and using reputation-based monitoring to catch illegitimate accounts and malicious activity, Judge says.

So far Twitter has mainly been hit by spam and phishing attacks, as well as hacked individual accounts. But a researcher at RSA Conference 2010 last week demonstrated a tool that impersonates a Twitter user's account in order to execute automated targeted attacks on the person's followers.

Pedro Varangot, a security researcher with Core Security Labs, says his team wrote the tool as a way to demonstrate and test for how social networks can be used for spear phishing.

Meanwhile, Twitter's Harvey said in his post that previously, his team was only able to detect phishing scams after the links had been sent. He called the new service "a major blow against phishing" and said that even if a malicious link is sent via an email notification and the recipient clicks on it, Twitter will "be able to keep that user safe."

Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.