Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

news.ycombinator.com

Post reply on HN
Print view View original

61–70 of 93 posts

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#61

Post by jcadam »

bpcrd wrote: post #11
candiddevmike wrote: post #8

Are you concerned that your product will inadvertently improve deepfakes? Suddenly you've given them a baseline that they need to be better than, and hackers love challenges. I predict this will turn into a constant arms race like AV or copyright protection, and I don't think this will work in the long run. IMO, KYC needs to go back to in person verification. Everything you can do digitally can be faked or impersonat…

In 2017 deepfakes were pretty crude, today the avg person can’t tell a real face from a deepfake generated on a 5 year old iPhone. We expect the tech to continue moving in this direction. So, similar to anti-virus, we're approaching this problem with an iterative, multi-model solution that can evolve with the threat.

Biggest fear is people being put in prison based on deepfake video evidence.

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#62

Post by jka »

rdrdg wrote: post #59
jka wrote: post #53

Hey - great and important idea. Have you discussed / looked into sampling environmental radio noise at various frequencies and locations and then interpolating samples of them within the video and audio itself at recording-time? (ideally along with some kind of near-unfalsifiable timestamp signals and/or device keys to confirm that "yes, this unique device was here at this time and the proof is within the pudding)

Currently, we are consider image, video and audio data types. This multi-modal idea sounds interesting none the less, thanks for sharing :)

Thanks! You're welcome. The noise signature could be included in-band, and wouldn't necessarily require multiple machine learning models to verify.

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#63

Post by bpcrd »

skeeter2020 wrote: post #32
candiddevmike wrote: post #8

Are you concerned that your product will inadvertently improve deepfakes? Suddenly you've given them a baseline that they need to be better than, and hackers love challenges. I predict this will turn into a constant arms race like AV or copyright protection, and I don't think this will work in the long run. IMO, KYC needs to go back to in person verification. Everything you can do digitally can be faked or impersonat…

If I was them I'd be concerned that it DOESNT move in this direction. Not a lot of money to be made selling one-and-done fake detection software, but a fortune available if they can convince clients a perpetual subscription is a core requirement.

We are optimizing compute requirements by isolating only the part of the media that requires scanning, so that we can pass along the cost savings to customers. At scale, the costs will low enough to scan all files just like anti-virus.

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#64

Post by bpcrd »

fxtentacle wrote: post #31
candiddevmike wrote: post #8

Are you concerned that your product will inadvertently improve deepfakes? Suddenly you've given them a baseline that they need to be better than, and hackers love challenges. I predict this will turn into a constant arms race like AV or copyright protection, and I don't think this will work in the long run. IMO, KYC needs to go back to in person verification. Everything you can do digitally can be faked or impersonat…

I came here to write that. Me and my friends have a competition to get a (obviously CGI) picture of my living room filled with gold coins past those "is that photo real" platforms. So far, none has stood up to the test. The worst one so far was TruePic, who even gave me a certificate of authenticity and then kept that certification PDF online on their website until they heard about us openly mocking them, and then th…

We're up for the challenge! Send us your images / videos (ask@realitydefender.ai) and we'll send you the results!

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#65

Post by ehnto »

jcadam wrote: post #61
bpcrd wrote: post #11

Earlier quoted context omitted.

In 2017 deepfakes were pretty crude, today the avg person can’t tell a real face from a deepfake generated on a 5 year old iPhone. We expect the tech to continue moving in this direction. So, similar to anti-virus, we're approaching this problem with an iterative, multi-model solution that can evolve with the threat.

Biggest fear is people being put in prison based on deepfake video evidence.

At some point it might make video evidence inadmissible. I imagine a tool like Reality Defender offers could be used in court to tackle this, but now Reality Defender itself needs verification. Like all tools used to verify evidence I suppose. Is a negative from Reality Defender enough to prove that a video is real, or is it only valid proof when it's proving it's fake?

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#66

Post by huynhhacnguyen »

rdrdg wrote: post #58
huynhhacnguyen wrote: post #37

First of all, congratulations on the lauch! Your description about the "model of models" and combining their scores is really intriguing. Detecting deepfake is an interesting topic on its own and apparently there are lots of use cases that I'm not even aware of, partly due to my limited knowledge in this subject. There are a few points of I'm curious about (Beware that the following questions can be very silly, comin…

Thanks for your questions. - The input to our models are image, video, and audio. Based on the model, we can use parts of the image (esp faces) or whole image. Yes, we also incorporate metadata for better detection. - It's a fair concern. As quality of generative media increases, so does the sophistication of detection. Since, we fully understand how generative media is created, it gives our the leverage to reverse e…

Thanks for the response, I'll look more into this. So what you're saying is we need to understand the methods to devise countermeasures to such methods, instead of being model-agnostic. That sounds like a grueling task and truly an arms race. Best of luck to the team!

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#67

Post by rockemsockem »

sigil wrote: post #36

I too wonder whether cryptographic signatures will be the long run solution to deepfakes. Can you outline why you don't think that will be the case? Here's an alt solution to argue against: 1. The necessary PKI gets bootstrapped by social media companies, where deepfakes begin to seriously threaten their "all-in-on-video" strategy, and simultaneously look like an opportunity to layer on some extra blue-check-style ve…

Not OP, but personally I'd love to see this sort of tech baked directly into smartphones and cameras. Your camera could have tamper-proof hardware that embeds signatures like you describe into the metadata of your video/photos while you take them. Sure, the user can separate it, but like you said over time people just won't trust photos/videos without this assurance.

Apps like snapchat that let you apply filters could also add additional metadata, going through more hardware verified operations provided by things like Apple's secure enclaves, saying something like "X filter applied over frames N-M" and you can have an edit chain to go with the original video/photo.

With this infrastructure in place our phones could really serve as ways to certify reality itself.

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#68

Post by rockemsockem »

asail77 wrote: post #56
sigil wrote: post #46

Earlier quoted context omitted.

That makes sense. But if signed AV takes off, the video someone else took of you and shared likely bears their seal. And audiences decide how much they trust that source – just like they look for a CNN / BBC / etc logo in the corner currently.

Great discussion. Many have asked the same questions. We hope a standard will be created and used by all digital content creation tools. But this will take time. And even then, bad actors will not be deterred. They will always find ways to create fake content and pass it as authentic. We want to be there to fight them every step of the way!

There is a standard for this actually. Hot(ish) off the press: https://c2pa.org/

Re: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform

#69

Post by DonHopkins »

fxtentacle wrote: post #31
candiddevmike wrote: post #8

Are you concerned that your product will inadvertently improve deepfakes? Suddenly you've given them a baseline that they need to be better than, and hackers love challenges. I predict this will turn into a constant arms race like AV or copyright protection, and I don't think this will work in the long run. IMO, KYC needs to go back to in person verification. Everything you can do digitally can be faked or impersonat…

I came here to write that. Me and my friends have a competition to get a (obviously CGI) picture of my living room filled with gold coins past those "is that photo real" platforms. So far, none has stood up to the test. The worst one so far was TruePic, who even gave me a certificate of authenticity and then kept that certification PDF online on their website until they heard about us openly mocking them, and then th…

Maybe your objections triggered their Mock Object Detector?

https://en.wikipedia.org/wiki/Mock_object

Post reply on HN