AI & Deep Fake – The Endless & Scary pit ?

This small piece is motivated by two videos I have watched recently by one of the channels I follow.
It is a photography channel, but of a different kind (Jchristina – )

Videos referenced:



Most of us, if not all of us has come up with cases of morphed images in our life – spread as ‘shocking news’. It could be someone hosting Indian flag in Pakistan or someone hosting US flag in North Korea and what not. Which has been busted in due course by fake news busters (thankfully there are quite a few around the world). But their work is getting tougher with each day, thanks to better ‘skills’ of not just the spreader of fake news but also software (AI). Anyways, when we see an outrageous photograph – the normal, sane people take it with a bit of scepticism. Some news channels even show videos with edited audio to implicate people in a crime never committed by them.

But what if you saw a person speaking something on a video ? Will you still not believe the video? Let’s be honest – very unlikely. But that is what is happening. There was Tiktok video of ‘Deep Tom Cruise’ with quite a few videos of AI Tom Cruise (you can see it in the 2nd video link). Now can you say that this is not Tom Cruise but a ‘virtual’ Tom Cruise? Pretty cool and fun right ? Pause and think for a moment. This is information technology and not restricted to anyone or any organization, there are ways to ‘get’ it and it will get better with each day for sure.

Now imagine a situation – your country is going for an election (Say US presidential elections or Indian Lok Sabha Election). Just before 2 or 3 days of the election a video is leaked in the internet where one or more of the candidates saying outrageous things. What will happen ? Do you think people will wait for someone to bust the fake video BEFORE you cast your vote? What if those leaders actually said those words and the videos are genuine? How will you know? For that matter, how will ANYONE know? And this video need not spread from your country – it can be spread by your ‘enemy’ country to destabilize your country. What if someone or some organization creates a surveillance video of a crime that was never committed by a ‘nosy’ reporter?

So, how to counter it ? There is AI software available which can detect such fake videos. But as you will see in the same video – it can’t track the ‘fake’ Barack Obama video. Now – people like Obama and Cruise are hugely popular public figures, AI has practically unlimited data to track any unnatural part of their videos. Yet it failed to catch the fake. What about the everyday Joe or Jane? The AI have practically no data to track the oddness. Basically it will be an endless race between two AIs with each winning one round alternatively.

If the world wasn’t already in a mess – its getting messier very fast! Now you can’t even differntiate between true and fake.

The line between truth and lie is becoming blurry – practically invisible!

One comment

Comments are closed.

%d bloggers like this: