Martin R. Albrecht
Martin R. Albrecht
@malb.bsky.social
Cryptography Professor at King's College London and Principal Research Scientist at SandboxAQ. Erdős–Bacon Number: 6. He/him or they/them. https://malb.io
941 followers309 following136 posts
Apropos of nothing, a few links on the "Spy Cops" scandal in the UK. - www.theguardian.com/uk-news/2023... - www.theguardian.com/uk-news/2022... - www.theguardian.com/uk-news/2021... - www.theguardian.com/uk-news/2020... - www.theguardian.com/uk-news/2020...
This is new. In eprint.iacr.org/2025/794 ("[3]" below) we reported the vulnerability that WhatsApp leaves group management to the server, cf. arstechnica.com/security/202... Now an "Let's build an AI backdoor" paper is like: "great stuff!"
Martin R. Albrecht reposted
Anthropic may not have intended to write an excellent ad for GLM 5.3, but that's what they did. www.anthropic.com/research/glm... Unlike Mythos, I might actually have a chance at using GLM for defensive research. Anthropic didn't reply to any of my requests for Mythos access for use on OpenSSH.
"Antifa: a howto and opsec guide" through the medium of song www.youtube.com/watch?v=Gg-S... genius.com/Genius-engli...
Martin R. Albrecht reposted
Martin R. Albrecht reposted
We finally finished the universal signature forgery for 1024-bit RSA! 2^32 oracle queries, 1200 core years precomputation, 180 core years for an individual forgery, and 3 years of human labor (no AI involved) by Laura, Adam, Nadia, Emmanuel and me to pull of this computation against real HSMs.
Forging 1024-bit RSA signatures in nearly SNFS time Hand over your HSM for some time, and we can forge signatures for its key. Arbitrary signatures. Forever. github.com/ucsd-hacc/NS...
Martin R. Albrecht reposted
We were fired by our accountant because apparently Cedarcrypt and my Lebanon course are money laundering operations: www.linkedin.com/posts/nadimk...
Martin R. Albrecht reposted
TL;DR: using compression before encryption is much weaker than we thought. Tiny length differences can be amplified - almost without limit - and noise-based or bucketization countermeasures are then easy to bypass. Joint work with @prefix-free.bsky.social and Lenka Mareková, to appear at CCS 2026.
Criminology: Refined Techniques for Compression Side-Channel Attacks (Yuanming Song, Lenka Mareková, Kenneth G. Paterson) ia.cr/2026/1972
Martin R. Albrecht reposted
Our latest “Crypto in the Wild” project, analysing secure email gateways. We give 29 cryptographic attacks on 4 different products, from SEPPmail, Cisco, Proton and CipherMail. Paper to appear at ACM CCS 2026.
Why Johnny Should Not Delegate Email Encryption to Gateways (Andris Suter-Dörig, Matteo Scarlata, Kenneth G. Paterson) ia.cr/2026/1911
Worth a read on who/what Autistici/Inventati is, the tech collective designated as a "Specially Designated Global Terrorist" by the US cavallette.noblogs.org/2026/08/10083
Martin R. Albrecht reposted
We also bought EncroChat's original domain (encrochat.ch) --- it now hosts our paper and other info.
Also, the article makes it sound as if our work is mostly concerned with the legal side of the Encrochat cases, but we gave a comprehensive technical account: eprint.iacr.org/2026/1319 We also discuss those open questions. We didn't have access to a device/image, limiting what we could establish.
As funny as it is that they simply dropped frida.re on Encrochat devices, to me the two big Encrochat questions that still aren't resolved are:
Revealed: How French cyber spies turned to GitHub to hack the EncroChat phone network used by organised criminals www.computerweekly.com/news/3666493...
Also, the article makes it sound as if our work is mostly concerned with the legal side of the Encrochat cases, but we gave a comprehensive technical account: eprint.iacr.org/2026/1319 We also discuss those open questions. We didn't have access to a device/image, limiting what we could establish.
As funny as it is that they simply dropped frida.re on Encrochat devices, to me the two big Encrochat questions that still aren't resolved are:
Revealed: How French cyber spies turned to GitHub to hack the EncroChat phone network used by organised criminals www.computerweekly.com/news/3666493...
OpenAI claim a proof that CPV is NP-hard for polynomial approximation factors. openai.com/index/ten-ad...
Martin R. Albrecht reposted
This is a very cool and exciting discovery by Claude Mythos! It found a serious 𝒎𝒂𝒕𝒉𝒆𝒎𝒂𝒕𝒊𝒄𝒂𝒍 attack on the post-quantum signature scheme HAWK, an "on-ramp" candidate for potential NIST standardization. www.anthropic.com/research/dis...
I recommend you read @nadim.computer's excellent write-up of why he teaches in Lebanon symbolic.software/blog/2026-07...
Our timeline really is Terminator: The Next Token Prediction openai.com/index/huggin...
A Real-World Law-Enforcement Hack: The Case of Encrochat martinralbrecht.wordpress.com/2026/07/01/a...
The abductions and torture continue in Kenya xcancel.com/KTNNewsKE/st... www.standardmedia.co.ke/national/art...
Hardness of hinted ISIS from the space-time hardness of lattice problems martinralbrecht.wordpress.com/2026/06/12/h...
Dear Journalists, That "crosshair" you're referring to is a Celtic cross, a well-known white supremacist logo: en.wikipedia.org/wiki/Celtic_... www.adl.org/sites/defaul... This is what these rioters are: white supremacists.
Martin R. Albrecht reposted
Security in Times of Surveillance eipsi.win.tue.nl/surveillance... w Marrtin Albrecht @malb.bsky.social Chloé Berthélémy (EDRi), Rikke Bjerg Jensen (RHUL), Thomas Lohninger @socialhack.bsky.social Björn Ruytenberg @0xiphorus.infosec.exchange.ap.brid.gy Carmela Troncoso @carmelatroncoso.bsky.social
Martin R. Albrecht reposted
Our statement on the UK gov't's demand that all content on all devices sold or used in the UK be scanned, on the presumption of nudity, using a dystopian combination of age verification & content scanning. This proposal will not safeguard children. It endangers us all. signal.org/blog/pdfs/20...
Leiden Declaration on Artificial Intelligence and Mathematics leidendeclaration.ai#declaration
arstechnica.com/security/202... with a quote by @bedow.bsky.social on how we saw/see no evidence of what they're alleging.
I'm reading some papers on AI safety (don't ask). Is the entire field really so naive that they think speaking of "human values" makes sense? Have these people not switched on the news ever? I know we computer scientists do not excell at thinking about the actual world we live in, but this is extra.
Martin R. Albrecht reposted
Replying to @ipv.sx
Similar to the E2E encryption debates in that nerding harder will not square the policy circle. Dissimilar to the E2E encryption debates in that the policy world is proceeding nonetheless, as @matthewdgreen.bsky.social said at the top.
‘They singled out non-white, foreign-born workers’: the restaurants raided by Britain’s version of ICE www.theguardian.com/uk-news/2026...
Fabio is looking for a PhD student in "Explainable, Knowledge-driven AI and ML for Systems Security" fabio.pierazzi.com/opportunity/...
Martin R. Albrecht reposted
The program for our Eurocrypt affiliated event CAW on May 10 is (mostly) finalized and published on: caw.cryptanalysis.fun We received many super exciting submissions! To register, select our workshop during conference registration on the Eurocrypt website once that opens: eurocrypt.iacr.org/2026/
Not Safe for Politics Cellebrite Used on Kenyan Activist and Politician Boniface Mwangi citizenlab.ca/research/cel...
Martin R. Albrecht reposted
Martin R. Albrecht reposted
Martin R. Albrecht reposted
Just finished presenting this work at Real World Crypto in Taipei :) TL;DR: We found 2 attacks on Signal (Android, Desktop) where a malicious server can inject messages in conversations. Super fun project! Thanks a bunch to Noemi Terzo, @kennyog.bsky.social, and @cryptojedi.bsky.social
Signal Lost (Integrity): The Signal App is More than the Sum of its Protocols (Kien Tuong Truong, Noemi Terzo, Kenneth G. Paterson) ia.cr/2026/484
Martin R. Albrecht reposted
I just donated to help equip Lebanon's first responders and firefighters with essential life-saving supplies to help them deal with the massive crises unfolding due to Israeli attacks on civilian areas. Please consider donating: fundahope.com/en/campaigns...
Fernando is looking for a PhD student www.iacr.org/jobs/item/4164 Fernando is excellent, you should consider applying.
Martin R. Albrecht reposted
Martin R. Albrecht reposted
A thread in which @sockpuppet.org presents some of the juiciest morsels from our paper at zkae.io :
"presenting a cornucopia of practical attacks". These are my favorite words ever to have occurred in a cryptography paper.
“Based on these ethnographic findings, we initiate the cryptographic study of at-compromise security” martinralbrecht.wordpress.com/2026/02/17/b...
At-Compromise Security: The Case for Alert Blindness (Martin R. Albrecht, Simone Colombo, Benjamin Dowling, Rikke Bjerg Jensen) ia.cr/2026/252
Martin R. Albrecht reposted
Hardness of hinted ISIS from the space-time hardness of lattice problems (Martin R. Albrecht, Russell W. F. Lai, Eamonn W. Postlethwaite) ia.cr/2026/187
Apropos of nothing, here's a piece on Reform's plans for a British ICE: shado-mag.com/articles/opi... Here's the policy document: web.archive.org/web/20260127... Here's a piece on the Labour Government's practice of emulating the US' celebration of brutality: www.theguardian.com/politics/202...
Martin R. Albrecht reposted
This is right. This is the message. This is the urgency. Minneapolis and so many other places are living under paramilitary occupation and tyranny by out of control, unaccountable federal thugs. This cannot stand. www.youtube.com/shorts/hyqtA...
Social Foundations of Cryptography: Autumn School London, UK | 15 to 17 September 2026 social-foundations-of-cryptography.gitlab.io/school