Software that can tell when you’re lying may be used to spot criminals at airports

FACIAL recognition software that spots when someone is lying could help weed out terrorists and drug smugglers at airports.

After analysing millions of frames of footage, researchers have discovered what they say are giveaway facial expressions of untrustworthy individuals that could be spotted by smart surveillance cameras to assess passengers.

The artificial intelligence would help officers decide whether a suspicious person in an immigration queue was telling the truth or just nervous. It could also minimise cases of racial profiling by staff. In the study, computer scientists at the University of Rochester, in New York, created the largest public data resource of liars’ expressions. In a few weeks, they amassed 1.3 million frames from 151 pairs of individuals. To collect the footage they used a system such as Skype, where one participant played the interrogator and the other lied or told the truth while being filmed by their webcams.

Using a machine-learning program that detects patterns in video footage, the researchers spotted new expressions that are associated with liars. Tay Sen, one of the scientists, explained: “It told us there were basically five kinds of smile-related ‘faces’ that people made when responding to questions.”

They found an intense smile involving cheek, eye and mouth muscles, called the Duchenne smile, was one of the most frequently associated with not telling the truth. Liars appear to “smile with their eyes” in the expression, which Mr Sen said is consistent with the “duping delight” theory that “when you’re fooling someone, you tend to take delight in it”.

Ehsan Hoque, a co-author of the research, said: “We still want humans to make the final decision. But as they are interrogating, it is important to provide them with some objective metrics.”