Applying new technologies to combat misinformation and its consequences on society
Disinformation is spreading on several fronts and its techniques are becoming increasingly sophisticated. It can affect the news we receive from the media, discredit institutions or individuals, manipulate elections or try to interfere in other public affairs of crucial importance to any democracy.
To combat misinformation, we must work on several aspects. First of all, we must educate in the media to provide consumers with the resources they need to know how to discern. In addition, we must take advantage of technological innovations that offer us tools capable of identifying signals that are imperceptible to the human eye. Finally, cooperation between different organizations, companies and institutions is necessary to advance and consolidate the fight against fake news and content.
Synthetic media are photo, video or audio files that have been manipulated by artificial intelligence. They are used to make it appear that certain people have said things they never said or that they have been in places they were not really in. They are created by a continuously learning artificial intelligence that outperforms conventional detection technology.
In this sense, among the new technologies to combat disinformation is Microsoft Video Authenticator. A tool that can analyze photos or videos and give a percentage or score of how artificially manipulated the file is. How does it work? Microsoft Video Authenticator works by detecting the blending limit of deepfake elements. In addition, it analyzes subtle fading or grayscale elements not detected by the human eye. This tool was created using a public dataset from Face Forensic++ and has been tested in DeepFake Detection Challenge Dataset.
On the other hand, synthetic media generation methods are expected to become more sophisticated. Therefore, more and more powerful methods must be sought to ensure authenticity.
Other innovations against disinformation
Different companies are working on new technologies capable of detecting manipulated content and ensuring authenticity. In the Microsoft environment, two components are at stake:
- A tool built in Azure that allows content producers to add certificates to pieces. These certificates are incorporated into the pieces as metadata.
- The other component is a reader that checks these certificates.