To get us started, could you briefly introduce yourself and Guardian Project.
Jack Fox Keen: My name is Jack Fox Keen, I use they/them pronouns. I am the data scientist at Guardian Project’s ProofMode application which I will be discussing. A little bit about me is that I have a background in math and scientific computing. Now, I am currently a Phd student at University of California, Santa Cruz, focusing on explainable AI in the Computer Science Department. Then I’m actually gonna share my screen to talk about Guardian Project because we have this really beautiful slide deck that is very exciting and we’re very proud of it.
So this is the Guardian Project and this is our team. We are a mobile and privacy technology group. We’re essentially a collective of hackers, developers, designers, activists, community leaders who are working to give people a voice, enhance safety, ensure authenticity, provide access to knowledge, and we’re open source free software. That’s been us from the get go.
The Guardian Project has actually been around for 15 years and it started off in 2004, over 15 years. We really got kicked off in 2009, focusing on network, chat and visual privacy. We have a couple of different apps called Orbot, Chat Secure, ObscuraCam, offering privacy protections. Then, in 2012, we started Verified Visuals, working with advocates and journalists on tools for verified, authenticated media. This is actually the birth of ProofMode. We also do off-gride communication things, basically trying to figure out a connectivity and low connectivity or no connectivity places such as focus on Internet shutdowns. For example, using our ButterBox. We also focus on physical safety, we have apps for helping journalists, LGBTQ communities, as well. Also, we are focused on respectful measurements. We have a thing called Clean Insights, which is a way of ethically collecting and analyzing data. Then we also have secure chat applications as well as we’re assisting organizations with defending news. In addition, with helping verify war crime evidence, and other human rights violations as well. That is the Guardian Project, we cover a whole breadth of things. I specifically work on ProofMode, which is the verified visual application. That has been my role, for since I’ve been with Guardian Project.
Data plays such a large part in society, and we often refer to it as objective but this isn’t always the case. Can you talk more about the ethics of data and how decentralized, specifically blockchain technology, can help with data ethics?
Jack: I think the most ethical thing that ProofMode does is that we actually don’t collect data on our end. All the photos that people take, for example, all the metadata associated with their photos, that is all in their hands. We see nothing of it. We really wanted it to be an app that we’re actually focusing on how to help people analyze their own data. Part of what my job has been, has been to work on this tool called ProofCheck, which is a decentralized app, a fully decentralized app. The nice thing about that is that it runs on your local server. It’s not running on a centralized server that is owned by us, it’s just running on your local server. Wih ProofCheck some of the things I’ve been working on is extracting the EXIF data from the photos in a way that a user could feasibly put that data into their own data analytics tool that they want. You know, Tableau or Glamorous Toolkit or something like that, same thing with the JSON data. I think the most ethical thing that one can do with data is to actually completely put it into the users’ hands of how and when they want to share it. If they ever decide they want to share it. Maybe they never want to share it. But that should be the user’s decision. Then, as far as the decentralized app, this might kind of cover a couple of other questions. We are working with the Inter-planetary File System (IPFS), and we’re thinking about decentralized storage of these verified visuals. There’s a line where you need to, as a activist, as a person, you need to decide; are you more concerned with censorship, or are you more concerned with privacy? That’s where decentralized storage systems has to navigate that spectrum. So if you have very sensitive photos, for example, and you’re more concerned about privacy, then maybe in that case a decentralized application, with all the potentials and mirrors, maybe wouldn’t be your best bet. But if you’re more concerned about censorship and you’re more worried about a photo being taken down from social media then, that’s where decentralized systems really play a powerful role in allowing you a robust way of storing a file that can’t just be censored by an authoritarian government, by social media companies. It’s a way of allowing the robustness of the photo existence, the evidence of that photo. The way that ProofMode works is, we have this hash, the ProofMode will basically create a hash of the photo as soon as you take it. The beauty of the hash is that a hash acts as a one-way function. So if I take a photo or take a picture, and I have that photo and then I have the hash of that photo, I can put that hash on the decentralized web and basically have multiple copies of that string of characters that’s identifying the photo. I’m also protected because it’s not like somebody can reverse engineer the hash to get the photo. You can input the photo into the hash function to get the hash and prove that that photo matches the hash. You can’t find a photo from the hash. For people who are dealing with really sensitive data, this is a really useful tool and then being able to replicate that hash and decentralized file system, or essentially storage system, is a good way of navigating that spectrum of privacy and censorship. I think this is something that I really want people to understand better, is that you do have a lot of privacy with the hash, but then you also have all the benefits of preventing censorship from centralized storage systems.
How is Guardian Project looking at and incorporating other technologies, such as AI, to support their mission?
Jack: With AI, our focus is how ProofMode is a tool to protect against the consequences of AI, especially with generative AI images. We know that there is going to be A) an influx of fake images and then B), there’s going to be an increase of people claiming that photos that they don’t like are fake, and we’re trying to figure that out. That’s what we want ProofMode to be able to be a guard against that we can say no, we have this in-app camera and when you take a photo with that in-app camera we know that its that photo. We have a Google Safety Check application that goes through the application as well. So if you have Google Safety Check verifying that you didn’t doctor your computer, that you didn’t alter it, or add a camera simulator, when you have that verified, and then you have the in-app camera verified and then all those different tools verified and all these different checks that go off. Then we’re trying to guard against claims that a photo was created by generative AI. At the same time, we also want to see how we can threat model our application by using generative AI to see here is a generative AI picture, or here is the worst case scenario of somebody submitting a fake data set or something like that. We’re using the adversary to make our tool more robust against it, this is the most prominent way that we are planning on working with AI. I think it’s a really good way of recognizing that there are these tools that can be used for both good and bad. And we’re seeing, how is it? How is the bad use case scenario able to make us stronger and how can we work, basically test our model against the worst case situation to figure out how we can make our app better and better and better.
If you could interview anyone, who would it be and what would you like to talk to them about?
Jack: As far as at this moment, at this point in time of what I’m focusing on, I actually would love to interview the authors of ‘Data Feminism,’ Catherine D’Ignazio and Lauren Klein. That is the focus that I want to be on, is thinking about how we can ethically work with data sets, how we can empower people. One of the big things about data feminism is citizen science and really working with communities, especially because there’s so much extractivism in data science. There’s a lot of white savior complex, this kind of extractive white saver complex with data science. I’m really excited about learning more, especially as I think about ProofMode, and even my own studies, how to go about data science in a way that is not extractive, and that we can work with the communities. Also, something I think about a lot is how we can compensate communities for their expertise and knowledge, because I feel like lived experiences are expertise, and people should be compensated for the knowledge that they’re sharing. That also comes back to funding and then how do you find ethical sources of funding. These are all big questions that I have as a data scientist, how do we actually create equity in these systems that are inherently inequitable. I feel like ‘Data Feminism’ is really starting to ask those questions so I would like to interview those authors.
But why blockchain?
Jack: We’re really trying to make the benefits of blockchain accessible to people who are not into cryptocurrency, for example, the most traditional use cases. For me, personally, I find this huge benefit of blockchain just by the way that everyone finds this benefit, the immutability of the links in the chain. The cryptographically verifiable understanding of these ledgers and that’s what I find really powerful. A use case of ProofMode that I’ve been exploring is how can we use timestamps, oftentimestamps.org to put timestamps on blockchain and then that creates an immutable timestamp. Then what is the power of this fourth dimension of the world of time. What I have actually been exploring is the use of the benefit of understanding timestamps in a chain of custody. In general ProofMode is trying to use timestamps as part of a chain of custody in evidence. We’ve traditionally focused on international affairs, human rights activism. I’ve been exploring this new use case of a more localized use case with domestic violence victims because in the process of going through a domestic violence civil court case you have to provide evidence of all the conversations that you had that were abusive. For example, you have to kind of grab those screenshots, and then there can be situations where people delete those messages, or they delete these comments on Facebook, or they delete those comments on social media. I realize that when you have this immutable timestamp verifying I had this message, this abusive message, from a year ago, and then I had this other follow up message from 6 months ago. There is this timeline of evidence of what I’m trying to explain. I think that is actually very powerful and that is something that blockchain very uniquely offers is the immutability of what is put on there.
Guardian Project is on Twitter. Follow them to stay up-to-date on their work.
