Facebook Inc. today announced that it’s teaming up with Microsoft Corp. and a number of universities to launch the Deep Fake Detection Challenge, a $10 million contest meant to spur the development of new methods to identify doctored videos. 

The initiative takes aim at the specific category of doctored videos known as deepfakes. It’s a relatively new phenomenon in which artificial intelligence is used to superimpose a person’s face and voice on another individual. Though deepfakes are often easy to spot today, researchers fear that bad actors could abuse future, more advanced versions of the technology for misinformation campaigns and other nefarious purposes. 

The Deep Fake Detection Challenge is Facebook’s answer to the threat. “Our hope is that by helping the industry and AI community come together we can make faster progress,” Facebook Chief Technology Officer Mike Schroepfer wrote in blog post announcing the contest. 

The company will supply a set of deepfakes created with the help of paid actors to give researchers a target against which to test their detection methods. Facebook plans to release an initial batch of videos at a scientific conference in October, while the contest itself will take place from December through spring of next year. 

Competitions have proven to be an effective means of encouraging scientific innovation. The U.S. government’s recurring DARPA Grand Challenge, for instance, played a formative role in the creation of the autonomous driving ecosystem. A more recent sister competition likewise run by DARPA inspired researchers to develop a computer capable of automatically uncovering vulnerabilities in software.

For Facebook, the Deep Fake Detection Challenge is likely driven by more than just academic considerations. As the world’s social network, the company faces the risk of its platform becoming a channel for deepfake creators to distribute fake videos. Facebook already faces criticism for misinformation on its platform and deepfakes will make moderation all the more difficult, which gives it a strong incentive to invest in better detection methods early.

The urgency of the issue is underscored by the company’s disclosure that it won’t will leave the task entirely to the academic community and pursue its own research. “Facebook will also enter the challenge but not accept any financial prize,” Schroepfer wrote in the blog post.

“Technology to manipulate images is advancing faster than our ability to tell what’s real from what’s been faked,” said MIT assistant professor Phillip Isola, whose university is among the seven backing the contest at launch. “A problem as big as this won’t be solved by one person alone.”

Image: Facebook