Facebook Develops New Method To Detect, Track Source Of Deepfakes Using AI-Generated Imagery

Deepfakes have received a lot of attention from the public worldwide. It has become a powerful tool to influence perceptions. It is also used by bad actors for spreading misinformation and fake news. The technology allows the editing of videos easily. Technology is getting better every day. The doctored video content appears in a way that they are real. To crackdown on such content, social media giant Facebook has developed a new method. The method can detect and track the source of deepfakes. It also involves reverse engineering. It analyzes AI-generated imagery. This lets Facebook identify the model that created the video. Facebook said that it has signed a pact with researchers at the Michigan State University to double down on deepfakes.

Facebook said that this is the first time when such a method has been developed to detect deepfakes. It is also the first time when the source can be detected. This will be useful in curbing the spread of misinformation on the platform. Facebook has come under severe criticism in the US and different countries for not doing enough to check the spread of misinformation and fake news. The new Facebook method comes in the backdrop of viral videos of Facebook CEO Mark Zuckerberg and Speaker Nancy Pelosi. Pelosi appeared drunk in the video. She had come down heavily on Facebook called the social media platform a shameful company.

Facebook said that deepfake software is extremely easy to customize. But the new method of image attribution can help identify videos’ generative model and trace the activity of bad actors. The company said that researchers can now obtain information about the model used for creating deepfakes using the new method. It said that the method will be useful in the real world. Facebook said that the method is a step further to deduce data about deepfakes. Earlier, Microsoft had launched a new tool to combat the spread of misinformation by spotting deepfakes. The tool identifies synthetic media that are manipulated by artificial intelligence.