Alex Hoover
LightNews LightNews
banner
axhoover.com
Alex Hoover
@axhoover.com
140 followers 410 following 45 posts
cryptographer at Stevens

axhoover.com
Posts Replies Media Videos
axhoover.com
I'm looking for PhD students to join me at Stevens!! Please spread the word to anyone looking for an opportunity right across from NYC to research:
* secure data structures (e.g., PIR/ORAM)
* cryptography for AI-related applications (e.g., watermarking, steganography)
Fully-funded PhD Students
iacr.org
November 13, 2025 at 7:44 PM
axhoover.com
Are there good resources to understand concrete limits of steganography in terms of the entropy, false positive rates, support size, etc of both the covertext distribution and the steganographic output distribution?

Ideally, something which uses something like cryptographic syntax too
August 8, 2025 at 6:44 PM
Reposted by Alex Hoover
natanael.bsky.social
Who could've known

(except for everyone)
Prompt Injection Tricks AI Into Downloading And Executing Malware
[wunderwuzzi] demonstrates a proof of concept in which a service that enables an AI to control a virtual computer (in this case, Anthropic’s Claude Computer Use) is made to download and execu…
hackaday.com
January 26, 2025 at 8:10 PM
Reposted by Alex Hoover
theatlantic.com
“When everything on the internet demands attention, paying attention to anything becomes impossible.” @yair-rosenberg.bsky.social unpacks one game creator’s quest to rescue the internet’s promise from its present:
The Worst Page on the Internet
Neal Agarwal distills digital life to its essence.
www.theatlantic.com
January 27, 2025 at 1:15 PM
axhoover.com
What's the best way to get used to iO tricks in proofs? Just read a bunch of papers?

I've gotten used to some Sahai-Waters ideas, but I feel far away from being able to use iO creatively to prove something new
January 3, 2025 at 3:46 PM
axhoover.com
Is this latest achievement really such an impactful tipping point? I feel like we shouldn't consider completely shifting policy approaches until we move beyond language models as AGI candidates
lawfaremedia.org Lawfare @lawfaremedia.org · Dec 24
"o3 fundamentally breaks this barrier, demonstrating an ability to synthesize new programs and approaches on the fly—a crucial stepping stone toward artificial general intelligence."
Kevin Frazier, @arozenshtein.bsky.social, and @petersalib.bsky.social on OpenAI's new model.
OpenAI's Latest Model Shows AGI Is Inevitable. Now What?
The question is no longer whether AGI will arrive, but whether we'll be ready when it does.
www.lawfaremedia.org
December 24, 2024 at 3:50 PM
Reposted by Alex Hoover
senykamara.com
If you’re curious about the design and analysis of encrypted algorithms and encrypted databases, I’m putting together a collection of resources at encryptedsystems.org
EncryptedSystems.org
encryptedsystems.org
December 3, 2024 at 4:02 PM
Reposted by Alex Hoover
eprint.ing.bot
Improved PIR Schemes using Matching Vectors and Derivatives (Fatemeh Ghasemi, Swastik Kopparty, Madhu Sudan) ia.cr/2024/1885
Abstract. In this paper, we construct new t-server Private Information Retrieval (PIR) schemes with communication complexity subpolynomial in the previously best known, for all but finitely many t. Our results are based on combining derivatives (in the spirit of Woodruff-Yekhanin) with the Matching Vector based PIRs of Yekhanin and Efremenko. Previously such a combination was achieved in an ingenious way by Dvir and Gopi, using polynomials and derivatives over certain exotic rings, en route to their fundamental result giving the first 2-server PIR with subpolynomial communication. Our improved PIRs are based on two ingredients: • We develop a new and direct approach to combine derivatives with Matching Vector based PIRs. This approach is much simpler than that of Dvir-Gopi: it works over the same field as the original PIRs, and only uses elementary properties of polynomials and derivatives. • A key subproblem that arises in the above approach is a higher-order polynomial interpolation problem. We show how “sparse S-decoding polynomials”, a powerful tool from the original constructions of Matching Vector PIRs, can be used to solve this higher-order polynomial interpolation problem using surprisingly few higer-order evaluations. Using the known sparse S-decoding polynomials in combination with our ideas leads to our improved PIRs. Notably, we get a 3-server PIR scheme with communication 2^(O^(∼)((logn)^(1/3))), improving upon the previously best known communication of $2^{O^\sim( \sqrt{\log n})}$ due to Efremenko. Image showing part 2 of abstract.
November 22, 2024 at 5:12 AM
Reposted by Alex Hoover
lawfaremedia.org
"For decades, technologists have been making the point that the strongest and best form of communications security is provided by end-to-end encryption; it is well past time for law enforcement to embrace its widespread public use," writes Susan Landau.
End-to-End Encryption Is a Critical National Security Tool
Law enforcement and national security officials have fought end-to-end encryption for decades—but the technology is more needed than ever. 
www.lawfaremedia.org
November 21, 2024 at 7:17 PM
Reposted by Alex Hoover
durumcrustulum.com
i started refactoring a bunch of code and then halfway through i changed my mind - i hadn't commited 💀
November 19, 2024 at 2:49 PM
axhoover.com
I just recently started cryptology.city!

I think it could be a useful resource for the cryptology-curious! I drew a lot of inspiration from the Complexity Zoo, which I have personally liked.

Happy to hear ideas on how to tailor the site for crypto though!
handle.invalid Cryptology City @handle.invalid · Jul 5
Cryptology City is up! cryptology.city

It is a wiki page to share cryptographic knowledge. I hope it will be a resource for people who are interested in cryptography, so they can learn about all the interesting results that are out there.

What would you like to see added? 🔐
cryptology.city logo
July 5, 2024 at 2:49 PM
Reposted by Alex Hoover
eprint.ing.bot
Enhancing Watermarked Language Models to Identify Users (Aloni Cohen, Alexander Hoover, Gabe Schoenbach) ia.cr/2024/759
Abstract. A zero-bit watermarked language model produces text that is indistinguishable from that of the underlying model, but which can be detected as machine-generated using a secret key. Unfortunately, merely detecting AI-generated spam, say, as watermarked may not prevent future abuses. If we could additionally trace the text to a spammer’s API token or account, we could then cut off their access or pursue legal action. We introduce multi-user watermarks, which allow tracing model-generated text to individual users or to groups of colluding users. We construct multi-user watermarking schemes from undetectable zero-bit watermarking schemes. Importantly, our schemes provide both zero-bit and multi-user assurances at the same time: detecting shorter snippets just as well as the original scheme, and tracing longer excerpts to individuals. Along the way, we give a generic construction of a watermarking scheme that embeds long messages into generated text. Ours are the first black-box reductions between watermarking schemes for language models. A major challenge for black-box reductions is the lack of a unified abstraction for robustness — that marked text is detectable even after edits. Existing works give incomparable robustness guarantees, based on bespoke requirements on the language model’s outputs and the users’ edits. We introduce a new abstraction to overcome this challenge, called AEB-robustness. AEB-robustness provides that the watermark is detectable whenever the edited text “approximates enough blocks” of model-generated output. Specifying the robustness condition amounts to defining approximates, enough, and blocks. Using our new abstraction, we relate the robustness properties of our message-embedding and multi-user schemes to that of the underlying zero-bit scheme, in a black-box way. Whereas prior works only guarantee robustness for a single text generated in response to a single prompt, our schemes are robust against adaptive prompting, a stronger and more natural adversarial model. Image showing part 2 of abstract.
May 20, 2024 at 3:20 AM
axhoover.com
I think proofs are great! Even from uncertain assumptions, reductions give us a better understanding of what maybe true and lets us build cool new things

If we were truly stuck waiting for proofs, cryptography wouldn't exist as it does today
May 1, 2024 at 8:19 PM
Reposted by Alex Hoover
eprint.ing.bot
Leakage-Abuse Attacks Against Structured Encryption for SQL (Alexander Hoover, Ruth Ng, Daren Khu, Yao'an Li, Joelle Lim, Derrick Ng, Jed Lim, Yiyang Song) ia.cr/2024/554
Abstract. Structured Encryption (StE) enables a client to securely store and query data stored on an untrusted server. Recent constructions of StE have moved beyond basic queries, and now support large subsets of SQL. However, the security of these constructions is poorly understood, and no systematic analysis has been performed. We address this by providing the first leakage-abuse attacks against StE for SQL schemes. Our attacks can be run by a passive adversary on a server with access to some information about the distribution of underlying data, a common model in prior work. They achieve partial query recovery against select operations and partial plaintext recovery against join operations. We prove the optimality and near-optimality of two new attacks, in a Bayesian inference framework. We complement our theoretical results with an empirical investigation testing the performance of our attacks against real-world data and show they can successfully recover a substantial proportion of queries and plaintexts. In addition to our new attacks, we provide proofs showing that the conditional optimality of a previously proposed leakage-abuse attack and that inference against join operations is NP-hard in general. Image showing part 2 of abstract.
April 10, 2024 at 3:56 PM
axhoover.com
Do users of signed languages use/have a concept similar to the spoken notion of puns? Like mixing or replacing similar signs as a joke

#linguistics
March 17, 2024 at 10:54 PM
axhoover.com
This was a cool paper to see posted! I’ve been curious about this problem for a while now, and it seems like it’s mostly solved
eprint.ing.bot ePrint Updates @eprint.ing.bot · Feb 27
Tight Indistinguishability Bounds for the XOR of Independent Random Permutations by Fourier Analysis (Itai Dinur) ia.cr/2024/338
Abstract. The XOR of two independent permutations (XoP) is a well-known construction for achieving security beyond the birthday bound when implementing a pseudorandom function using a block cipher (i.e., a pseudorandom permutation). The idealized construction (where the permutations are uniformly chosen and independent) and its variants have been extensively analyzed over nearly 25 years. The best-known information-theoretic indistinguishability bound for the XoP construction (due to Dutta, Nandi and Saha~[IEEE Trans. Inf. Theory]) is about q²/2^(2n), where q is the number of queries and n is the block length. The XoP construction has also been recently analyzed in the multi-user setting and the best known bound for it (by Chen, Choi, and Lee~[CRYPTO’23]) is about $\sqrt{u} q_{\max}^2/2^{2n}$, where u is the number of users and q_(max) is the number of queries per user. A generalization of the XoP construction outputs the XOR of r ≥ 2 independent permutations, and has also received significant attention. In this paper, we improve all previous bounds obtained in the literature for the (generalized) XoP construction when q > 2^(n/2) (assuming q < 2^(n)/2). Specifically, for the basic XoP construction with r = 2, we obtain an indistinguishability bound of q/2^(1.5n) in the single-user setting and $\sqrt{u} q_{\max}/2^{1.5n}$ in the multi-user setting. Hence, if q_(max) = Θ(2^(n)) (and q_(max) < 2^(n)/2), then our bound of $\sqrt{u} q_{\max}/2^{1.5n}$ implies that the adversary’s advantage remains negligible as long as u = o(2^(n)). On the other hand, with the previous bound of $\sqrt{u} q_{\max}^2/2^{2n}$, the adversary’s advantage may already be a constant with a single user (u = 1). For the generalized XoP construction, we obtain a bound of q/2^((r−0.5)n) in the single-user setting and $\sqrt{u} q_{\max}/2^{(r - 0.5)n}$ in the multi-user setting. Consequently, the gap between our results and the best previous ones increases sharply with r. For example, the best-known bound for r = 3, obtained by Choi et al. [ASIACRYPT’22] (in the multi-user setting), is about $\sqrt{u} q_{\max}^2/2^{2.5 n}$, while we obtain $\sqrt{u} q_{\max}/2^{2.5 n}$. Since all of our bounds are matched (up to constant factors) for q > 2^(n/2) by attacks published by Patarin in 2008 (and their generalizations to the multi-user setting), they are all tight. We obtain our results by Fourier analysis of Boolean functions. Yet, most of our technical work is not directly related to the analyzed cryptosystems. It rather involves analyzing fundamental Fourier properties of the density function associated with sampling without replacement from the domain {0, 1}^(n). We believe that this analysis is of broad interest. Image showing part 2 of abstract. Image showing part 3 of abstract.
February 27, 2024 at 10:25 PM
Reposted by Alex Hoover
eprint.ing.bot
Plinko: Single-Server PIR with Efficient Updates via Invertible PRFs (Alexander Hoover, Sarvar Patel, Giuseppe Persiano, Kevin Yeo) ia.cr/2024/318
Abstract. We study single-server private information retrieval (PIR) where a client wishes to privately retrieve the x-th entry from a database held by a server without revealing the index x. In our work, we focus on PIR with client pre-processing where the client may compute hints during an offline phase. The hints are then leveraged during queries to obtain sub-linear online time. We present Plinko that is the first single-server PIR with client pre-processing that obtains optimal trade-offs between client storage and query time for all parameters. Our scheme uses t = Õ(n/r) query time for any client storage size r. This matches known lower bounds of r ⋅ t = Ω(n) up to logarithmic factors for all parameterizations whereas prior works could only match the lower bound when $r = \tilde{O}(\sqrt{n})$. Moreover, Plinko is also the first updateable PIR scheme where an entry can be updated in worst-case Õ(1) time. As our main technical tool, we define the notion of an invertible pseudorandom function (iPRF) that generalizes standard PRFs to be equipped with an efficient inversion algorithm. We present a construction of an iPRF from one-way functions where forward evaluation runs in Õ(1) time and inversion runs in time linear in the inverse set (output) size. Furthermore, our iPRF construction is the first that remains efficient and secure for arbitrary domain and range sizes (including small domains and ranges). In the context of single-server PIR, we show that iPRFs may be used to construct the first hint set representation where finding a hint containing an entry x may be done in Õ(1) time. Image showing part 2 of abstract.
February 26, 2024 at 3:09 AM
Reposted by Alex Hoover
lance.fortnow.com
STOC Accepted Papers

http://acm-stoc.org/...
February 19, 2024 at 10:32 PM
Reposted by Alex Hoover
anildash.com
www.ftc.gov/news-events/...
FTC Says Ring Employees Illegally Surveilled Customers, Failed to Stop Hackers from Taking Control of Users' Cameras
The Federal Trade Commission charged home security camera company Ring with compromising its customers’ privacy by allowing any employee or contractor to access consumers’ private
www.ftc.gov
February 12, 2024 at 2:36 AM
Reposted by Alex Hoover
lance.fortnow.com
We start off the favorite theorems of the last decade with Babai's blockbuster algorithm for Graph Isomorphism.

blog.computationalco...
February 7, 2024 at 3:51 PM
Reposted by Alex Hoover
lance.fortnow.com
Who wins in a perfectly played game of Othello? Nobody.

Paper by Hiroki Takizawa
arxiv.org/abs/2310.1...

Blog post
blog.computationalco...

Perfect Play:

November 9, 2023 at 1:49 PM
axhoover.com
I'm shocked that the more educated you are, the more likely it is that you're a privacy doomer

I feel like education should empower people with a better understanding and more alternatives! But it seems like this is lacking with privacy
www.pewresearch.org/internet/202...
Figure from pew research linked in text
October 18, 2023 at 10:04 PM
axhoover.com
cryptosky repost
malb.bsky.social Martin R. Albrecht @malb.bsky.social · Oct 12
I'm looking for a postdoc to work with us on lattice-based cryptography

See martinralbrecht.wordpress.com/2023/10/12/p... and www.kcl.ac.uk/jobs/076525-...

Closing date: 31 January, 2 year contract in London, salary £42k to £60k.

Please help me reach potential candidates.
Research Fellow/Research Associate in Cryptography
www.kcl.ac.uk
October 12, 2023 at 1:17 PM
axhoover.com
I feel like Cricket and Baseball don't make great Olympic sports, but was breakdancing really a better pick?
nytimes.com The New York Times @nytimes.com · Oct 10
Cricket, baseball, softball, flag football, lacrosse and squash have been proposed by the organizers for the 2028 Olympics in Los Angeles.

The International Olympic Committee will vote on the new proposals later this month.
A Plan to Shake Up the Olympic Menu Includes Cricket
Flag football, softball, squash and lacrosse would also be included in 2028 under the proposal.
www.nytimes.com
October 10, 2023 at 4:43 PM