Zemana Is Ready To Fight The Spread Of Deepfakes

Via Zemana.

A little known cybersecurity firm in the Balkans has joined the counter-current against the rise of deepfake video clips. Deepfakes began grabbing headlines after a wave of disruptive elections in 2016 when voters around the world had their social media accounts sown with fake news. Now that the post-facts era has begun in earnest a new threat has since been identified. The easy utilization of video altering software that superimposes composite facial features on real subjects has given rise to concerns that both politicians and public figures can be “hacked” for spreading misinformation.

Enabled by deep machine learning algorithms that crunches large samples of a target person’s visual data, the resulting synthetic audio and imagery poses extreme risks for an unassuming public wired to social media. For Zemana, which has offices in Bosnia and Turkey, a multi-platform software developer kit called the Deepfake Detection SDK can now help anyone identify and filter inauthentic content. It works as a first line of defense against malicious video clickbait where, as Zemana puts it, “deepfakes swap celebrities’ faces into porn videos and put words in politicians’ mouths.”

Currently in its first alpha version Zemana’s homepage allows visitors to run video clips their software can analyze. The company warns of deepfakes becoming so accessible “it will soon be as easy to create  a fake video as it is to add an Instagram filter.” Zemana unveiled the Deepfake Detection SDK at the GTEX 2019 technology show in Dubai last month. According to a press release from Zemana their new software “enables developers to add a multilayered detection engine to a range of platforms…[to] enable governments, social media sites, instant messaging apps, and news organizations to detect AI-made forgery…before they cause harm to society.”

Zemana’s Vice President for Business Development Samir Mujovic considers deepfakes a global threat, describing it as “the first serious punch from artificial intelligence to humanity.” The promotional literature surrounding Deepfake Detection SDK is just as grim and acknowledges how “the science of detecting deepfakes is effectively an arms race.” Zemana was founded in 2007 and gained traction after the initial success of its malware scanner. Zemana now offers a modest product range of malware detection tools and endpoint security for businesses. Developers are encouraged to sign up for the Zemana newsletter for testing Deepfake Detection SDK and keeping tab on its updates.

At this point the growing concern over deepfakes is still emanating from the US where election security, along with mitigating cyber risks to American businesses, is an increasingly daunting prospect. Awareness of deepfake videos went viral in 2018 through sites like Youtube and seedier platforms for viewing porn. To date, the best known countermeasures for deepfakes are coming from researchers and small companies like Zemana. Of course, others in the cyber field have seen opportunity amid the crisis and deepfake detection is now a marginal trend among software developers. Another company that is using its expertise to mitigate deepfake media is Deeptrace from the Netherlands and the US government is also involved with creating tools against the same.

A distinction needs to be made, however. Using artificial intelligence to alter and edit video is a harmless activity. The problem arises in the current climate where political news is always distorted and once cherished editorial standards have crumbled. A nightmare scenario for malicious deepfake media is if it triggers a conflict or distorts financial markets, with both having potential global consequences.