RIT to develop software to detect deepfake videos
The $100,000 grant for RIT comes from the The Ethics and Governance in A-I Initiative, a joint project from the MIT Media Lab and Harvard's Berkman Klein Center for Internet & Society. This project and the other projects funded by this effort are aimed at helping combat the spread of misinformation, particularly when that happens through the use of artificial intelligence.
At RIT, researchers will work on designing software to help automatically detect what they call, 'deepfake videos.'
Matt Wright, an RIT professor of computing security who is also director of the Center for Cybersecurity at the Golisano College for Computing and Information Sciences, says RIT wants this software to be able to detect something like a fake presidential address that someone might put up on social media.
"It's really important for journalists to understand, is this real, should I put it out on our website, should we report on it?" Wright said. "Or is it fake, and then what do we do? And that's up to the journalist, but to understand more about the context of what's going on."
Wright said these 'deepfake' videos can have serious implications for society. "They could have significant social impacts as well as potentially making people feel like, 'oh, maybe I shouldn't believe anything in the news or any other sources because anything could be fake.'"
RIT researchers hope to develop this software later this year and WXXI News journalists will help them beta test some of that software.