 Hello, my name is Peter Hall. I am a PhD student at New York University, and in this video abstract for my joint work with Daniel Aepan, Chloe Cachet, Benjamin Fuller, and Feng Hao Lu, on our work Non-Maluable Digital Lockers and Robust Physi extractors in the Plain Model. Now, we're going to break down that title term by term, first starting with point obfuscation and then as an extension digital lockers. Point obfuscation is essentially obfuscation of the underlying point program that you see on your screen, which is kind of instantiated with some point x, and then should output one if and only if the testing point x prime is exactly equal to x. In this work, we're considering and constructing VBB obfuscation, and specifically we're trying to do this in the Plain Model. We also want non-malleability, which as we discussed in the full talk is actually pretty subtle to define, but abstractly means that given an obfuscation of a point x is hard to get an obfuscation of a point x prime that is some way related to that point x. Additionally, we're interested in digital lockers, also known as multi-bit point function obfuscations, which essentially, instead of outputting one or zero, output this multi-bit string or output perp. There's been much effort in recent years on Plain Model non-malleable point obfuscation. This line of work was initiated by Komar Gadski and Yoga. However, none of these works are actually sufficient for constructing entirely non-malleable digital lockers, point function obfuscations again with multi-bit output. However, we do use assumptions that are rooted in Bartosek-Mann-Zandri's crypto 2019 paper, which they introduce and justify there. Digital lockers can be constructed from composable point obfuscations that are useful in many settings. In this work, we're considering one application of them, specifically robust fuzzy extractors. Fuzzy extractors are initialized with an enrollment string from a kind of noisy source, and they should return some stable random value when queried on other values that are close to this original enrollment. The robustness guarantee, in addition, shows that from being given a fuzzy extractor, one cannot tamper that fuzzy extractor to a new one that outputs a different random string instead of the original one that was produced at enrollment. Prior works have constructed robust fuzzy extractors in a variety of ways, as you can see from this table here, but there's actually been no plain model constructions that can handle input sources with very low entropy, in this case less than half their length. Our contributions are as follows. We define a new object called non-malleable point function obfuscations with associated data. Intuitively, they try to leverage the non-malleability of the point obfuscation to attest to the correctness of a CRS-like string, which we call the associated data. We construct this above, and then we additionally achieve composability with some other obfuscation primitives, in particular, regular point function obfuscation. We show how the associated data is sufficient to use alongside an appropriate non-interactive zero-knowledge system to construct non-malleable digital lockers. And finally, we use syndrome decoding with all of the above to achieve more powerful robust fuzzy extractors in the plain model, specifically that are able to remain secure over low entropy inputs. This kind of represents a little diagram of the roadmap, and you're welcome to come to the full talk if you want to see how we do this along the way. Finally, just to once again bring it back around to robust fuzzy extractors, one application of these non-malleable point function obfuscations with associated data is we use them to achieve two constructions of plain model robust fuzzy extractors that remain secure when enrolled from sources with entropy less than half their length. Thank you, and I hope to see you in the full talk.