Is he a psychopath? Just look at his Tweets.

While psychopaths, often described as people who are superficially charming and extremely intelligent, are very hard to spot in the real world, they are easily exposed virtually on Twitter, scientists claim.

Researchers found that a person's tweets can reveal whether they are a psychopath, especially if they often use words like "die", "kill" and "bury."

A person who frequently swears may also be a psychopath, according to computer science researchers at the London's Online Privacy Foundation.

The scientists claim that word choice can indicate personality traits and that the latest findings could be used by authorities to identify potential threats or by employers to make hiring decisions.

However, while the latest study shows how social media can assess a person's personality it can potentially be used to falsely label someone as something they're not.

"People are making judgments about others based on social media. Companies even exist that will do this for you if you're hiring," said Chris Sumner, who heads the privacy foundation, according to the Sun Sentinel. "However, almost all research says more research is needed before social media screening should be considered for use."

Researchers said that while previous studies of a person's language and mental health have been conducted before, few have looked at social media, where a person can easily share his or her thoughts and feelings with the world.

Researchers Randall Wald, a doctoral student in engineering and computer science, and Taghi Khoshgoftaar, a professor of engineering and computer science at Florida Atlantic University, used an existing psychological formula to determine how likely a person is a psychopath based on their writing.

Using a computer program to scan tweets and looking at questionnaire answers researchers revealed that 1.4 percent of the 3,000 participants had psychopathic tendencies.

"It's one indicator [of psychopathy]," Wald told the Sun Sentinel. "It's not going to give you 100 percent accurate results. It's not enough to send in the SWAT team because someone is highly rated on this."

Researchers noted that there were some limitations to the study. For instance, the computer program didn't recognize abbreviated versions of words, which frequently happens on Twitter because of the 140-character limit for tweets. The program also could not recognize the difference between someone using a word like "kill" in an angry way or in a joking say like "I could kill him for this."

Researchers plan on presenting their findings in December at the International Conference of Machine Learning and Applications in Boca Raton, Florida.

RELATED STORIES