Microsoft recently announced that it has built a new video technology to help combat the rise of deepfakes and in turn, help with the increasing disinformation problem on the internet. This announcement comes at a crucial time when the U.S. is heading into a tumultuous election, that's also taking place in the middle of a pandemic.

Of late, there has been increased attention on fighting disinformation and misinformation online. This can be traced back to the impact that targeted mass disinformation campaigns had on the 2016 U.S. presidential election and the UK's Brexit referendum. Social media platforms, especially Facebook and Twitter, were severely criticized for not regulating the content posted to them. During the pandemic, misinformation of various kinds have surfaced and have even had a very real impact on people's behavior. This is also one of the reasons why social media platforms have been employing AI tools and making policy changes to regulate content that is likely to contain misinformation. Twitter has, at least in the U.S., been extremely proactive in dealing with misleading and harmful content, especially by high profile handles, including President Donald Trump. Facebook, although slow to respond and seemingly less keen to regulate, has also reformed its content policies, in light of recent pressure.

Related: Trump Campaign Reportedly Doctored Facebook Ad Footage To Age Joe Biden

Now, it seems Microsoft has also stepped in to help with its new Microsoft Video Authenticator that's designed to detect deepfakes. The company announced the new technology solution in a blog post and explained how it will be an integral part of it's Defending Democracy Program, along with the voting process protection tool ElectionGuard that was launched earlier this year. The video authenticator has been jointly developed by Microsoft Research with the company's Responsible AI team and the Microsoft AI, Ethics and Effects in Engineering and Research (AETHER) Committee.

How Microsoft's Video Authenticator Detects Deepfakes

The video authenticator analyzes photos and videos to a give a percentage chance or a score on the likelihood of them being artificially manipulated. The tool can give percentages for each frame of the video in real-time as it plays. This is done by detecting the blending boundary between manipulated images and the greyscale elements hardly visible to the human eye. The AI tool was created by training the algorithm using the popular training models for deepfake detection technology, Face Forsensic ++ and the Deepfake Detection Challenge Dataset. Built into the Microsoft Azure AI ecosystem, the tool lets content producers add a marker that travels with the video and can be read by the video authenticator to produce a degree of authenticity, represented by a 'confidence score.'

The technology will power the BBC's news verification technology 'Project Origin' and will also be part of the Trusted News Initiative launched by the several major international new media groups. Microsoft emphasizes that the tool will be updated to keep up with the technology that helps the creation of synthetic media. Microsoft's development and focus on the disinformation detection technology has partly been influenced by the research conducted by Professor Jacob Shapiro of Princeton University. Shapiro's findings included that 96 separate foreign influence campaigns had targeted 30 countries between 2013 and 2019.

Mass disinformation campaigns are prevalent on social media today and are capable of swinging public opinion on key issues and events. Due to this influence, it is extremely important to deal with this issue since democracy and sensible debate depends on healthy information channels. In this sense, technologies focused on combating misinformation, including Microsoft's latest, will be crucial in ensuring honest information makes its way to people.

More: Facebook Won't Crack Down Election Ads Until The Final Week

Source: Microsoft