Real-time deepfakes can cause lasting financial damage

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

1. 1:26

   I want to first start with the deepfakes because I'm super interested in that. What's the problem right now? Like, can you just tell folks the scale of the problem when it comes to deepfakes, Zoom? Like, where are we? Yes. So I think probably all of us at this point have like read some story on the news or on Twitter about some CFO, some employees, maybe some people that you know that they got a call pretending to be someone and they asked them for money or they told them something and they acted on that.

2. 1:55

   And then afterwards, it turned out he was a deep fake. And I've heard this everywhere from executives that we work with, like really big firms that wired like millions and millions of dollars to like my uncle who like got a deep fake and thought he was my cousin. Yeah, I've had like 20 stories of someone who gets a DM on Telegram from an account that's hacked, but it actually is that person. And saying, this happened to DC, I think I'm allowed to share that.

3. 2:17

   His Telegram got hacked. And then often they send a link to like, oh, we should catch up a Calendly link. And then when you join the Calendly link, it's an actual, it's them, but a deepfake version of them. So it's like a friend in their actual account and then a deepfake version of them. Like that's freaky. Exactly. And this is just exploding because just recently we crossed a threshold where models can now create video of more than 50, of more than 60 FPS frames per second in real time.

4. 2:43

   And so what that means is that you used to, for deepfakes, have to like pre-generate them and then use them. It was complicated, but now you can in real time be generating these like really sophisticated images. And so, of course, many industries, but especially the financial industry, are really concerned about this, right? Like imagine, for example, one of our design partners for this is OpenExchange.

5. 3:01

   And OpenExchange is maybe not very well known for most people, but it is in the financial space because what they do is they facilitate all the earnings calls from public companies. So you can imagine what happens if a deepfake version of a CEO gets on an earnings call, shares some news, that moves stock price. It doesn't matter whether five minutes later or two hours later you realize that was a deep fake the stock already moved right and millions of dollars uh have gone and moved and like even companies can blow up just for that right and so uh all over the financial ecosystem this is something that people are incredibly concerned about whether it's internal conversations conversations externally and so we're starting with like high risk individuals right with the executives and the ceos and all of them you can imagine that as ai continues to

6. 3:44

   advance you want that to actually trickle down throughout the whole organization so we've partnered with Zoom. We are actually in Zoomtopia the next two days with them in Los Angeles, launching this together with their CEO. So we're super excited to bring that. And of course, like everything in the world is like fully privacy preserving. Yeah. That's super cool. You guys are launching with Zoom. I'm glad that you talked about the importance around the financial use cases.

7. 4:05

   I think when you hear deep fakes, of course, no one wants deep fakes, but you think a lot of the consumer use cases. You don't really think about the earnings call and how that might actually move markets. What's the gravity when it comes to financial use cases and how AI is coming to play and changing just the scale of the problems? Paint that picture. What's going on? Yeah. So I think that, first of all, there's just a scale of the problem.

8. 4:29

   Even whether you're talking about bots on the internet, Cloudflare just announced that now more than 50% of the traffic online is bots. I expect that to be 99.999% very soon will actually be bots, not humans. And so just whether you're a bank trying to trust KYC when a user is going through your flows, you're a company doing internet calls just across the board. The thing about the financial ecosystem is that the stakes are much higher.

9. 4:53

   right if i do a deep fake of you pretending to do something funny i post it on twitter it may be embarrassing for you but then you can go back and apologize and there's the impact may be smaller and but there's a little less of a permanence the thing about the financial ecosystem is you're talking about like people's livelihoods and you're talking about like really big impacts and it's actually really hard to roll back the changes like what we were talking about right you you move the the stock price with with earnings call and whatever the impact what happened like that money is gone right or if you get defrauded that money is gone and so it's just really important that we solve it there and then yes in the rest of the internet