← All talks

Unmasking The Deepfake Threat: Detection, Prevention, And Navigating The Future

BSides London · 202513:46163 viewsPublished 2025-02Watch on YouTube ↗
Speakers
Tags
StyleTalk
About this talk
Deepfakes have become an increasing source of concern as AI advances. These extremely lifelike, digitally made videos can be used to propagate falsehoods, harm reputations, and even commit financial crimes. This talk would go into the complexities of deepfake technology, discussing how it is made and the potential repercussions. We will talk about effective detection techniques, preventive measures, and the role of legislation in tackling this increasing problem. Understanding the issues offered by deepfakes allows us to better navigate the digital realm and protect ourselves from their negative consequences.
Show transcript [en]

Good afternoon, everyone. Good afternoon, everyone. Um today I'll be speaking about unmasking the deepfake traits. We'll be speaking about the deepfake. We'll be speaking about prevention and also we'll be speaking about navigating um the future of deepfake. Um before we go um just a brief disclaimer, I'm not an expert in this field. Um if you want to correct me, you could do that when we go out. Um secondly, I could be nervous. I could lose true say in Yoruba, too. Please encourage me by laughing. Ah, I have a cold. I'm a cold. If I sneeze in between, please pardon me. And also I don't have the rights to all the way images I might be sharing with you

today. Thank you very much. Um before we go, I would like to We're talking about deepfake and I would like I thought about introducing myself with deepfake. Um I don't know how this might sound. Good afternoon. My name is Introduction. This afternoon I'll be speaking to you on the topic of deepfake. I'll be speaking to you about what deepfake is all about. I'll be speaking to you about the misuse of deepfake. I'll be speaking to you how to spot deepfake. I'll be speaking to you on how to prevent being a victim of deepfake. And also I'll be talking about navigating through the future of deepfake and AI in general. I hope we'll learn a lot in this class together. Sit

back and relax. Thank you very much. Let us begin. Um who knows that person? That's Stormzy, but that's not Stormzy. That's me speaking. So that's just deepfake. Stormzy is one presenting, but that's my voice. Um what is deepfake in general? Deepfake um we can say deepfake is uh images in which the person in the the existing picture or the image or audio has been altered using AI technology. Um deepfake is um derived from two words. Um one of them is deep, the other one is um fake. The deep part is um an AI language, a subset of AI which is deep learning. While the other one is fake, which indicates that um it is a

misleading content. Um deepfake has become um um an increased source of concern as AI advances. Um this extreme lifelike technical or digital videos can be misleading and also can propagate falsehood. It can cause harm to a petition and also it can cause to it can lead to suicide and also for to financial crimes. Also um the big question is this, are deepfake illegal? Is deepfake illegal? Um it depends on where you're looking from look looking at it from. Deepfake could be legal and it might be legal. Um it depends on the intent, it depends on the use and the jurisdiction on which is being used for. On the other side, um deepfake could be used for educational

purposes, for entertainment and even in movies and in musical videos. We can see in Kendrick Lamar's video, he did a video whereby he changed from um from Drake to Kanye West. He changed to about three to four different people in that video. It could be used for entertainment purposes. It could be used for movies. We saw in Fast and Furious 7 where it was used for Paul Walker that was late already. We saw him appearing in Fast and Furious 7. That was deepfake. It could be used for good purposes and also it could be used for bad purposes. In the UK, under the um online safety under the online safety act um which was passed last year,

the sharing of deepfake was made illegal and a new law is currently making its way in the parliament that will make it offensive and a crime for someone to create any explicit video or content of anyone without the person's permission. Um we're going to look at the dark side of deepfake. But what we're going to be looking at the actors, we're going to look at look at the dangers and also we're going to be looking at the misuses of deepfake. Number one, who are the actors involved in deepfake? We see there are state actors actors. These are people that are probably financed by the government or a big institution. Also we can see fake news outlets.

We see the blogs carrying fake news. We can see cyber soldiers. We see cyber criminals. We see fraudsters. We see hackers. Um also what could be the damage of the misuse of deepfake? Number one, we see misinformation. We see false news. Um deepfake when deepfake can fabricate false narratives by making people appear to say or do things they never did. By you see some political um people figures coming out to endorse something they never said. Um this could lead to misinformation. It could lead to fake news. We see earlier last year when the fight between Ukraine and Russia where the president came out and said he was calling off the war. That was misinformation. That was an

element of deepfake. Also it could cause political unrest, it could cause social unrest. Um also deepfake could lead to identity theft and um fraud. Um many people today use deepfake to scam relatives. Some of them use it to scam big organizations. Also it could lead to damage of reputation. I've seen some um celebrities whereby their videos are leaked online which never they never did and it causes them the endorsement deal. Nobody wants to really associate with them. So this can damage people's reputation. Also it can be used for extortion. It could it could be used for blackmailing. It could be it could be used also for online um abuse and harassment. Um how can we The next

question would be, how can we spot deepfake? Um number one, the easiest way to spot deepfake is sometimes you unnatural facial expression. That's Barack Obama and that's my face on it. And let me try to play it and see if it makes makes any sense. I do have one final ask of you as your president. The same thing I asked when you took a chance on me 8 years ago. I'm asking you to believe. Not in my ability. I do have one final ask of That doesn't look like me totally because you can see some unnatural facial feature. You can see my um beard not really jointed together. You can see my face not moving

properly. Um with all these you could easily spot um deepfake. And also these are now one awkward looking body of posture. This is Stormzy. This is my cool picture. I made it um Stormzy Stormzy's look like me, but look at his body. His body it doesn't have a build up like me. Look at his neck. His neck does not appear in the picture. I don't know if you guys can see this part. So um you see awkward body um um body or postures. These are the ways you could easily spot um deepfake. Also sometimes I look at this, I wish I'm this handsome. This is me and Beyoncé. I hope Jay-Z won't be jealous. Um sometimes you see strange colors. Um

if you see my lips in here, you could see different colors. If you see the beards, you could see um different colors in the beards. Even down the beards here, you could see the lower part is a bit darker while the upper part is a bit gray. These could be areas where we could easily spot what is deepfake and what is not deepfake. But looking at it, it looks so real. It looked like I was chilling with Beyoncé. I wish I was. Um another place is um blurry edges. If you could look at my neck here, you could see it's a bit blur. But go to the picture. If you look if you just look at it, you look

how it's so nice. But if you look at look at it deeply, you could see a bit blurriness at the edge. These are signs that um the images are deepfake. Um a bit of disclaimer, I've never been to a PDD party. No offense. Um this is a another picture of me looking so handsome with Beyoncé and PDD. Um another way of us spotting deepfake is reverse image search through Google or any image search um engine. Look at it. I ran it through Google and we could see a lot of pictures of for the of the resource where it comes from. You can see who is really there, not me um but um Jay-Z. Um also

I don't know if you noticed when I played the video, the voice was actually mine. The voice was actually Barack. The The face was mine. You could see sometimes the audios and the noi- you you could see the audios might not tally with the face of the person. The voice of the person might be quite different. Also you can notice sometimes some form of noise in the video. That could be another area you could easily spot if it's a deepfake or not. Um another way we could easily spot when something is deepfake if it's not featured in one of the most trusted media houses. Take for instance, we don't see BBC talking about the news. We

don't see I I ITV talking about the news. We don't see Channel 4 news talking about the news. You don't only see it on blogs. You only see it on WhatsApp status. We shouldn't rely on them. They might be fake. Most times they are fake. If they're not on trusted big boss newspaper channels and TV station. And also after all this, the next thing is protecting yourself or ourselves from deepfake. How do we protect our self from deepfake? Um, while deepfake present a challenge, there are steps we can use to protect our self. Number one, we have to be critical. Uh, we we don't take everything we see online with face value. You do not see

any picture, you see video, "Oh, this has happened. Someone said this. This is somebody's nude picture." No, we shouldn't just take everything we see online with face value. We should look for signs of manipulation. We should look for natural movement like all what I've tried to show to you guys. We should look at the lightning, we should look at the shadow, we should look at some other things in the video before we believe what we see. Number two, we should also verify um, information before we pass it on. We should not rely on just one single source of information, especially um, in this area of fake news. We should always cross reference, cross examine, and also look

at where the source is coming from. We should not just um, take anything we see. We should Sometimes if you if you can't confirm it, go ask some any of your friend that you know that is well enlightened that is exposed, and maybe is a tech savage. Also, we should we should be beware of malicious or suspicious links. Sometimes we get some links that have been sent to us, and we click on them. It could lead us to somewhere with our where we we could be scammed, where our identity could be used for deepfake. Also, we should always stay informed. I think this is a key awareness in everything we do. We should always be

aware. We should stay informed. We should know that deepfake is evolving. Sometimes you see some videos, you see some images, they appear more real than the main one. We should always be aware that technology is evolving. And also, the best way to spot deepfake presently is that there are some other deepfake tools like sensitivity. I'm not posting them here because they are not sponsoring me. We have some some other AI tools we could always use to cross-check if an image is fake or not. So, lastly, navigating through the future. The future of deepfake detection inches on a multi-face approach that combines technology, innovation, and ethical consideration, and public awareness. By addressing these challenges and seizing these opportunities, we can

work together towards a future where deepfake can effectively be detected and their harmful impact will be minimized. Um, before we go, I will just want to share some images for fun. Um, we know who is that? Neymar and Mbappe. Um, cool. What best? Um, Bobby? Um, You know who is this?

You don't know who is this? We don't have Madrid fans in here? Vini? Vini Jr.? Uh, what about that? Yeah. Um, then the last person, Foden. Yeah. Um, that's all for me. Thank you all for listening to my BORING JOKE.

I WANT TO SHARE UM, BEFORE I GO, THANK YOU ALL. BEFORE I GO, A SPECIAL UM, shoutout to my thank you to my mentor Jessica and my beautiful mentor also, Nikki Webb. She has mentored me all this while. Jessica was also useful um, in mentoring me, reviewing this with me. I thank you all. Thank you very much for coming.