Experts warns of deepfakes in hands of miscreants

Deepfakes, or videos or images that look surprisingly real, have already created ripples in the entertainment industry, but they also pose huge dangers to countries like India with a diverse population as miscreants could easily use these manipulated media to create rift among different communities, warn experts.

For the uninitiated, deepfakes are hyper-realistic media created with the help of Artificial Intelligence (AI). They show people saying or doing things that they did not say or do.

These digitally altered videos are far more dangerous than texts and regular photoshopped images carrying false or misleading information as much more people are likely to trust them.

"Deepfakes pose a special danger to democracies with large populations and lower literacy, especially those with a history of communal sensitivity," leading tech policy and media consultant Prasanto K Roy told IANS.

"While images and videos have been engineered for yore, sometimes for harmless fun, deepfakes posit a real threat to the social fabric. Imagine a video that could look like a real evidence of thoughts and intents of prominent authorities and celebrities," added Prabhu Ram, Head, Industry Intelligence Group (IIG), CyberMedia Research (CMR).

In the hands of bad actors, such technology can be a really simple tool to spread misinformation, Ram said.

The DeepNude deepfake app that was shut down two weeks ago after experiencing massive traffic, allowed users to virtually "undress" women with a few clicks.

How real deepfakes appear could be easily gauged from the digitally altered video of Facebook CEO Mark Zuckerberg that went viral on Instagram earlier this month.

In the video, deepfake Zuck, as the character is being called, appears to be speaking to CBS News in which he was seen saying: "Imagine this for a second: One man, with total control of billions of people's stolen data, all their secrets, their lives, their futures."

"Nice redirect campaign, exposing 'deepfakes', when we all know this is the real truth," wrote one user on Instagram after watching the video.

"OMG! Tech has never been better... Way to go!" wrote another.

While the Facebook-owned photo-sharing app did not take down the video, Zuckerberg is reportedly considering to treat "deepfakes" and misinformation separately so that they could be tackled in a better way.

But experts believe that fighting deepfakes would be even tougher than tackling other forms of fake news.

According to a Pew Research Centre report released this month, most adults in the US believe altered videos create a great deal of confusion about the facts of current issues and events.

They also believe that the public should not be expected to know when a video or image has been altered or made up entirely.

So with deepfakes a situation can arise when people cannot believe what they see with their own eyes. On the other hand, they provide bad actors with what may appear as "evidence" to make false claims about some people or communities.

"Combating such disinformation campaigns can be really tough, and AI researchers are still out-surpassed by deepfakes. There are some automatic detection systems, but they are too early stage," Ram said.

"Deepfake tools can be banned here and there but they will be accessible. But some steps such as efforts to enhance tech and resources to identify and flag deepfakes could help," Roy added.

Social media companies should work with fact checking organisations as well as on technologies to flag such content as fake because it may not be possible to intercept, identify or take down messages or media content in current, encrypted messaging platforms such as WhatsApp or Telegram or Signal, he added.