How Hugging Face Is Changing Who Gets to Build Artificial Intelligence

I'm here with Irene Suleiman, Chief Policy Officer and SVP of Strategy at Hugging Face. Thank you so much for being here, Irene. Thanks, Justine. Of course. So talk to us, first of all. I think Hugging Face has been in the news enough that most people know what it is and what's going on over there. But for people who don't, just give us a brief summary of what is happening with Hugging Face right now, where we're at. Well, we've had a busy summer, and it's kind of wild to be a household name from that little friendly emoji. Think of us sort of like a public library for all your favorite AI artifacts, models, and data sets. And so when people think of Hugging Face, they might think about some of the things that have been in the news recently. Talk to us a little bit about that, just in general, what we're seeing and how this is kind of in this entire AI landscape right now. Most relevant for the everyday person is the cyber attack that we defended over the summer. So there was an agent swarm from OpenAI that we were notified was attacking our infrastructure and we had to respond quickly with open models. Yeah, and thinking about your position now there, working with policy, working with strategy. How does your position and your experience really help when it comes to these situations? Having a multidisciplinary background to be able to inform the public, policymakers about an unprecedented incident is really important for calming people down as panic does not forget good decision making. And also having a technical background and understanding the appropriate responses is really urgent. Yeah, that balance is so important to be able to juggle all of that and have kind of a cohesive response to everything and being efficient and effective. And thinking about what happened and also now thinking about moving forward, what are you seeing from changes you're making, things that you're working on within the company to kind of help in this situation? I mean, most urgently now is around cyber defense. We hear use AI for cyber defense, but it's quite unclear to the everyday person not only how to do that, but what to deploy and what level of compute you need. We're trying to translate better and also just give more tooling and availability to the everyday person that might not be able to afford or get access to frontier technology. Yeah. And when you think about that technology and also the kind of governing agents here, there's a lot going on in the AI space when it comes to governance, when it comes to regulation. What are you seeing in that space and how do you think that's kind of affecting this in general? There's a lot of concern around the concentration of power. And we also just see a lot of hunger among everyday researchers, but also enterprise for a more lightweight, affordable, and open models. Yeah, and so when we talk about moving forward, there's obviously a lot of potential regulation coming down the line. A lot of different companies are calling for different types of regulation. What do you think should be done? Again, maybe not just at Hugging Face, but just in general, when it comes to moving forward in this AI kind of new world that we're in. For regulation generally, transparency does go a long way. Building better standards, there are already good research protocols for doing that. codifying that could really give some calmness to the discourse. In regards to cyber, incident reporting like we did in July with the cyber incident is just a really easy step to move forward. Yeah, that's such an important step to at least have the transparency and let people know, okay, this is what's going on or not. When it comes to people who are sitting out there and saying, well, now I'm scared to use AI or now I'm scared to go on Hugging Face, what do you say to those people who are just worried at this point? It's understandable to be worried and worry, panic is not going to help us. It's not very helpful to say that. But what can we do, I guess, is the response that I want to ask. What can we do to make sure that we feel secure and we're making sure that folks have the level of defender access they can, the level of understanding of technology so that we can be more informed? And thinking about that kind of fear that people have, a lot of different fears when it comes to AI, but one of the chief concerns is privacy. And then, if there is any hack, but also how the actual company is using their data anyway. So what do you say to those people who are specifically focused on this fear of what's happening with all of my personal information? I mean, often privacy is a trade-off with convenience. So if you have important IP, personal data, this is where local models go a really long way. I host my own small, what we call quantized models on my own laptop, on my MacBook. In terms of what you're willing to share, I think it's, is it easy? And are you willing to make that trade-off with something that has a good user interface? Or do you want to scale up? And there are a lot better resources for learning how to have more privacy-preserving AI. Yeah, that balance, again, always so important to think about what do you actually want to be sharing or not. It's so important. As we move forward to the rest of the year and also looking a little bit down the line the next few years, what do you think is going to happen next when it comes to this world of AI that we're living in and all of the different companies kind of working together and some together, some against each other within the ecosystem? I mean, I'm a dreamer. I'm hopeful. I see a world of open and closed. I think we just have a lot in common across the AI ecosystem and a lot of interest in making sure this goes well and goes safely. I see a balance of how everyday people are deploying both open and closed, even in enterprise. So I see the world going a little bit better than I think the sentiment might feel. Well, I appreciate the positivity and thank you so much for taking the time to speak with us. I really appreciate it. Thanks.