Fighting fake news
With nearly every major news story comes a barrage of fake photos spread across the internet, often reaching viral status on social media. But it can be challenging and time-consuming to verify which images are authentic and which have been Photoshopped, preventing people from getting accurate news and skewing the online conversation.
Undergraduate students Rohan Phadte, an electrical engineering and computer sciences major, and Ash Bhat, an interdisciplinary studies major, figured there had to be an engineering solution for this problem. They’ve created a machine learning tool that helps people identify when an online photo has been doctored or is fake news. Called SurfSafe, their browser plug-in instantly checks online images against news sources and fact-checking sites, such as Snopes.com and Factcheck.org, then places warnings on falsified images. Just one month after its release, SurfSafe already had 4,000 users.
“In this day and age, we should all be reading fact-checking sites, but it’s so unreasonable to expect each of us to spend 30 minutes online every day reading these things,” said Bhat. “Instead, we’ve created an extension that you can download in 30 seconds and that will do all of the fact-checking work for you.”
SurfSafe is just the latest offering from the duo’s startup, RoBhat Labs, which they launched out of their apartment in 2017. Last fall, they released BotCheck.me, a browser extension that uses advanced machine learning to detect and tag posts from political propaganda bots. One year after its release, BotCheck.me has more than 62,000 users and boasts an accuracy rate of 97.5 percent.
So what’s next? Phadte and Bhat — friends since middle school — have now brought five other Berkeley undergrads onto their team. Their plan is to make RoBhat Labs “the Norton AntiVirus of fake news,” using machine learning techniques to fight online misinformation, in all its insidious forms.