The feature was first Monday by Mikko Hypponen, chief research officer with security company F-Secure. When someone tries to post a link to a malicious Web site, Twitter pops up a short notification saying "Oops! Your tweet contained a URL to a known malware site," and, after a few seconds, deletes the post.

F-Secure says it's recommended that Twitter start doing this because the site "is increasingly targeted by worms, spam and account hijacking," according to Hypponen's blog post. A month ago, technology entrepreneur Guy Kawasaki's account was misused to post a link to a malicious Web site. In recent weeks users have been hit with links to fake, and sometimes malicious, "rogue" security software.

Security experts said Monday that while Twitter's filtering is a good first step, it still needs some work.

In tests, the feature blocked a URL that led to a phishing site, but it allowed the same link to post if it was shortened using services such as Tinyurl.com or Bit.ly. Because Twitter enforces a strict 140 character limit on each message, these URL shortening services are the most common way of posting links to Twitter.