Matthew Green
Cryptographer; teaches cryptography at Johns Hopkins and writes A Few Thoughts on Cryptographic Engineering.
Matthew Green did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest x.com
What they believe
-
Guarding a sandboxed AI model with a weaker monitoring model leaves the alignment problem unsolved, because the guard model still has to be trusted.
a warden-guarded sandbox is just another version of the alignment problem. You're going to have to trust a model to do it, and that model will need to be at least some fraction as intelligent as the model it's guarding.
blog.cryptographyengineering.comSaid 30 Sept 2026This on korrents
-
Over the next two years, major pieces of software are likely to run out of remotely-exploitable bugs.
Thus: over the next two years, major pieces of software are likely to run out of remotely-exploitable bugs.
blog.cryptographyengineering.comSaid 14 Aug 2026This on korrents
The bold line is korrents’ wording; the quote is theirs. These 2: what they wrote before what they said aloud, one per source, newest first.
Feed
As its own page →Hiding
3 November
2 November
30 October
29 October
28 October
27 October
11 October
30 September
From one piece Is sandboxing sufficient to contain rogue agents? 3 beliefs · blog.cryptographyengineering.com
-
Their words
Forget the swarms and the super-intelligence. What OpenAI really learned this summer is much worse: its agents will do what they're told by whoever manages to get text in front of them.
-
Their words
you can't perfectly isolate agents, at least not if you expect them to do useful things. Agents are most useful when they have access to information.
-
Their words
a warden-guarded sandbox is just another version of the alignment problem. You're going to have to trust a model to do it, and that model will need to be at least some fraction as intelligent as the model it's guarding.
14 August
From one piece Everything is about to “go dark” 2 beliefs · blog.cryptographyengineering.com
-
Their words
The destruction of the low-hanging vulnerability fruit will make law enforcement (and intelligence) agencies' need much more acute. The demand for constructed, intentional backdoors will re-start in earnest.
Everything is about to “go dark”blog.cryptographyengineering.com
-
korrents.com
Over the next two years, major pieces of software are likely to run out of remotely-exploitable bugs.Their words
Thus: over the next two years, major pieces of software are likely to run out of remotely-exploitable bugs.
Everything is about to “go dark”blog.cryptographyengineering.com
29 July
9 June
29 May
6 May
17 April
23 March
2 March
2 February
23 January
25 December 2025
13 November 2025
29 October 2025
15 October 2025
14 October 2025
10 October 2025
2 October 2025
27 September 2025
10 September 2025
6 July 2025
2 April 2025
31 March 2025
Nothing matches.
Beliefs
Korrents What they believe 16 beliefs — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
Guarding a sandboxed AI model with a weaker monitoring model leaves the alignment problem unsolved, because the guard model still has to be trusted.
a warden-guarded sandbox is just another version of the alignment problem. You're going to have to trust a model to do it, and that model will need to be at least some fraction as intelligent as the model it's guarding.
Is sandboxing sufficient to contain rogue agents? Said 30 Sept 2026
AI agents cannot be perfectly isolated if they are to be useful, because their usefulness depends on access to information.
you can't perfectly isolate agents, at least not if you expect them to do useful things. Agents are most useful when they have access to information.
Is sandboxing sufficient to contain rogue agents? Said 30 Sept 2026
The deeper danger of AI agents is that they do what they are told by whoever gets text in front of them, not that they are superintelligent.
Forget the swarms and the super-intelligence. What OpenAI really learned this summer is much worse: its agents will do what they're told by whoever manages to get text in front of them.
Is sandboxing sufficient to contain rogue agents? Said 30 Sept 2026
Show 13 more
Over the next two years, major pieces of software are likely to run out of remotely-exploitable bugs.
Thus: over the next two years, major pieces of software are likely to run out of remotely-exploitable bugs.
Everything is about to “go dark” Said 14 Aug 2026
As AI-driven bug hunting removes easy software vulnerabilities, law enforcement and intelligence agencies' need will grow more acute and demand for intentional backdoors will restart in earnest.
The destruction of the low-hanging vulnerability fruit will make law enforcement (and intelligence) agencies' need much more acute. The demand for constructed, intentional backdoors will re-start in earnest.
Everything is about to “go dark” Said 14 Aug 2026
Telegram is not an encrypted messenger in the sense that matters, because it does not end-to-end encrypt conversations by default.
Telegram clearly fails to meet this stronger definition for a simple reason: it does not end-to-end encrypt conversations by default.
Is Telegram really an encrypted messaging app? Said 25 Aug 2024
Almost every one-to-one Telegram chat, and every group chat, is readable on Telegram’s own servers.
The practical impact is that the vast majority of one-on-one Telegram conversations — and literally every single group chat — are probably visible on Telegram’s servers, which can see and record the content of all messages sent between users.
Is Telegram really an encrypted messaging app? Said 25 Aug 2024
Encryption people have to switch on protects almost nobody; what protects people is what a service turns on for everyone.
To be honest though, it doesn’t matter how secure something is if people aren’t actually using it.
Is Telegram really an encrypted messaging app? Said 25 Aug 2024
Telegram urges people away from default-encrypted messengers while refusing to build the features that would encrypt its own users.
Indeed, it no longer feels amusing to see the Telegram organization urge people away from default-encrypted messengers, while refusing to implement essential features that would widely encrypt their own users’ messages.
Is Telegram really an encrypted messaging app? Said 25 Aug 2024
Default cloud backup has quietly put a copy of nearly everyone’s private papers in a few data centres, and that is a surveillance weapon.
It terrifies me, because these data repositories are not only a risk to individual user privacy, they’re effectively a surveillance super-weapon.
Why encrypted backup is so important Said 7 Dec 2022
Nobody chose cloud backup: a handful of Silicon Valley executives made the choice for everyone in pursuit of adoption metrics.
A handful of Silicon Valley executives made the choice for us, in pursuit of adoption metrics and a “magical” user experience.
Why encrypted backup is so important Said 7 Dec 2022
A tracing requirement in an encrypted messenger gives up the confidentiality of who sent what, which is the part that protects the sender.
In short: traceability can really screw with the “who sent what” side of content confidentiality.
Thinking about “traceability” Said 1 Aug 2021
Who sent what is not metadata, so an encrypted system that reveals it has given up more than the routing information it could never hide.
Information revealed about “who sent what” in an E2E system is not the same as metadata.
Thinking about “traceability” Said 1 Aug 2021
Content tracing works by turning every user into a cooperating witness, whether or not they want to cooperate with the police.
This brings us to the central challenge of all content tracing proposals so far: to make tracing possible, a tracing system needs to turn every WhatsApp user (including the originator) into a cooperative green circle — regardless of whether users actually want to cooperate with police.
Thinking about “traceability” Said 1 Aug 2021
Shrugging that perfect security is impossible is the wrong lesson: the achievable goal is to wreck the economics of mass exploitation.
The problem that companies like Apple need to solve is not preventing exploits forever, but a much simpler one: they need to screw up the economics of NSO-style mass exploitation.
A case against security nihilism Said 20 Jul 2021
What made NSO different was not its exploits but its scale: it turned state-grade surveillance into something any government could buy.
Rather: NSO’s genius is that they’ve done something that attackers were never incentivized to do in this past: democratize access to exploit technology.
A case against security nihilism Said 20 Jul 2021
Beliefs others hold too
Encryption people have to switch on protects almost nobody; what protects people is what a service turns on for everyone. 2 hold this
To be honest though, it doesn’t matter how secure something is if people aren’t actually using it.
Is Telegram really an encrypted messaging app? Said 25 Aug 2024
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
Recommends rcmnd
rcmnd What they recommend 5 things — each backed by an exact quote.
Compiled by rcmnd.app, not by them: the quotes are theirs, and each links to the post it was copied from.
Recent
Loved Modern Cryptography
Modern Cryptography by Wenbo Mao. An excellent introductory crypto textbook, a little bit less dense than the HAC.
blog.cryptographyengineering.com 12 Jan 2012
Loved Foundations of Cryptography
Foundations of Cryptography by Oded Goldreich. This two-book pair is probably the best cryptographic theory textbook. This is a great book to start with if you're looking to understand the formal underpinnings first. But it's not an applied crypto textbook.
blog.cryptographyengineering.com 12 Jan 2012
Loved Introduction to Modern Cryptography
Introduction to Modern Cryptography by Jonathan Katz and Yehuda Lindell. We use this as a core text at Hopkins. A fantastic mix of practical and theoretical crypto.
blog.cryptographyengineering.com 12 Jan 2012
Show 2 more
Recommends The Handbook of Applied Cryptography
The Handbook of Applied Cryptography (aka the HAC), by Menezes, van Oorschot and Vanstone. One of the fundamental textbooks in this area. Covers basic theory, symmetric and asymmetric cryptography, and protocols. This is not a quick read, mind you. It's a serious textbook and an excellent reference.
blog.cryptographyengineering.com 12 Jan 2012
What others say about The Handbook of Applied Cryptography →
Recommends Security Engineering
Security Engineering by Ross Anderson. If you're designing or analyzing security systems and haven't read this book, you need to do so immediately. Anderson is a cryptographer, security expert and all-around savant.
blog.cryptographyengineering.com 12 Jan 2012
Where they publish
Blog A Few Thoughts on Cryptographic Engineering His blog on cryptography, encryption policy and what is broken this week. Has a feed.
Open blog.cryptographyengineering.com
Recent
- New website: estimating human cryptanalytic work 11 Oct 2026 This is not so much a full post as it is an advertisement. I’ve been working with Claude and GPT 6 Astra to conduct a survey of historical cryptanalysis results, in an effort to determine how many human-hours have been…
- Is sandboxing sufficient to contain rogue agents? 30 Sept 2026 Quick caveats: this is a post on AI safety, written by a cryptography professor. If that troubles you, you should read something else. I try hard not to work on AI (except when the topic occasionally tosses itself in my…
- Everything is about to “go dark” 14 Aug 2026 I’m coming down from spending a few days at Usenix Security, right here in my hometown of Baltimore. This means that my days have been taken up with two kinds of conversation: first, explaining to colleagues why Baltimo…
Show 7 more
- Some thoughts about Anthropic’s new cryptanalysis results 29 Jul 2026 Yesterday Anthropic published two new cryptanalysis results, both outputs of Claude Mythos, their (still) unreleased advanced model. The first of these results attacks a signature scheme called HAWK, while the second is…
- The future of Siri, or: why private inference isn’t private enough 9 Jun 2026 Yesterday Apple announced a big step towards deploying real AI in their Siri ecosystem. In most ways this is good and inevitable: Siri is one of the world’s most widely-used voice agents, and it would be good if it didn…
- Let’s talk about encrypted reasoning 29 May 2026 Update August 11, 2026: A group of researchers from all over Europe were inspired by this post, and actually turned it into a real working attack! Check out their writeup and paper here. This is a quick post I wanted to…
- Anonymous credentials: an illustrated primer (Part 2) 17 Apr 2026 This is the second in a series of posts about anonymous credentials. You can find the first part here. In the previous post, we introduced the notion of anonymous credentials as a technique that allows users to authenti…
- Anonymous credentials: an illustrated primer 2 Mar 2026 This post has been on my back burner for well over a year. This has bothered me, since with every month that goes by, I become more convinced that anonymous authentication the most important topic we could be talking ab…
- WhatsApp Encryption, a Lawsuit, and a Lot of Noise 2 Feb 2026 It’s not every day that we see mainstream media get excited about encryption apps! For that reason, the past several days have been fascinating, since we’ve been given not one but several unusual stories about the encry…
- Kerberoasting 10 Sept 2025 I learn about cryptographic vulnerabilities all the time, and they generally fill me with some combination of jealousy (“oh, why didn’t I think of that”) or else they impress me with the brilliance of their inventors. B…
Link verified 20 Sept 2026. Recent items update automatically from the channel.
x.com @matthew_d_green Posts.
Recent
- The way real Homebrew gets installed is so sketchy I guess downloading it from a Google ad isn’t totally unreasonable. 3 Nov 2026
- It’s pretty awesome for security researchers that we have a permanent MITM channel for inspecting unpatched IoT device traffic, thanks to eMuhdra. Pretty crap for everyone else, though. Would be worth thinking about how to do things differently in the future; I doubt this will be the last broken CA. 3 Nov 2026
- Apparently 1/3 of the global Danfoss refrigerator population is flagged as honeypots by Shodan 😂 3 Nov 2026
Show 17 more
- I care more about the fridges than the Binance hack because I’m offended that someone would destroy good cheese. 3 Nov 2026
- And the fridges got popped again. 2 Nov 2026 theguardian.com
- It is incredibly unhelpful that Anthropic framed their research on agent worms in terms of IoT and “model escape”. 30 Oct 2026
- I’m happy to see security people pushing back against some of the bad past decisions by AI labs. I also think security and sandboxing is critical (and I’m not a doomer!) But there’s also an attitude that new model capabilities are “no big deal” that sometimes gets packaged with that, and I very much don’t agree with that. 30 Oct 2026
- Cryptocurrency is like the technical heat shield for society. It’s a big pile of money that attackers will burn before they go after everything else we care about. 29 Oct 2026
- “First they came for matrix multiplication And I did not speak out Because Bitcoin did not use matrices Then they came for Navier–Stokes And I did not speak out Because Bitcoin does not use fluid mechanics Then they came for the unit-distance conjecture And I did not speak out Because Bitcoin did not use geometry Then they came for asymptotic SAT solving And I did not speak out Because Bitcoin was not asymptotic Then they came for elliptic curves And I was like oh shit” - @EranTromer 29 Oct 2026
- A couple of folks have pointed out a claimed asymptotically subexponential algorithm for ECDLP today. This is based on the ETH results and is very clever. But has zero practical implications whatsoever. 29 Oct 2026
- If your vulnerability policy is “only take exploits with execution roots seriously and close others that are PoC as informational” then I think researchers should feel entitled to just dump their non-critical findings publicly. 29 Oct 2026
- I threw Abliterated GLM at the crypto behind Telegram secret chats, and I think it killed itself. 28 Oct 2026
- Yes, the ETH is false. But it isn’t *that* false. In this series of tweets I will 27 Oct 2026
- So this is a silly story, but for a year or two before in 2013/2014 I went around telling everyone I could find about this little cryptography library called “OpenSSL” and how it was used everywhere, but had almost no developers or security budget. 27 Oct 2026
- We haven’t seen a significant cryptanalysis result from a frontier AI lab since late July. 27 Oct 2026
- How many of you had “the exponential time hypothesis is false” scheduled for October? I expected we had til Thanksgiving before stuff got really weird. 26 Oct 2026
- Weird how they’ve found so many amazing things and yet not one single result related to practical cryptanalysis. I guess everyone should pat themselves on the back for a job well done. 26 Oct 2026
- I don’t know how seriously to take this, but it’ll be pretty funny if every security researcher in the US gets pwned using open-weight models, because the frontier US labs wouldn’t give them access to their red-team models. 26 Oct 2026
- Someone on BlueSky took me to task about a post I wrote about Metamask’s cryptography a few years ago. And I’m kind of a masochist for that kind of criticism because I usually think I’m wrong or being too mean. 26 Oct 2026 blog.cryptographyengineering.com
- Exploit development is one of the most frustrating components of computer security, not because it’s bad (I am in awe of people who can do it) but because too many vendors make working exploits a requirement for patching things. 25 Oct 2026
Link verified 20 Sept 2026. Recent items update automatically from the channel.
Mastodon @matthew_d_green@ioc.exchange Posts.
Recent
- There’s something ominous about the speed with which the entire world has marched to require identification on platforms and, as I expected, begin the process of banning anonymous VPNs. 6 May 2026
- I’ve been working on a little simulator for MeshCore, written entirely using Claude Code. I was hoping some MeshCore devs might take a look and let me know if there are features they could use. https://github.com/matthewdgreen/meshcore_sim #meshcore 23 Mar 2026
- Microsoft is handing over Bitlocker keys to law enforcement. 23 Jan 2026 forbes.com
Show 12 more
- I was stupid enough to buy this new AppleCare One plan for a phone I bought my daughter. Now I learn this only covers the device if it’s connected to the same Apple ID (not family plan). Have to spend Christmas unwinding this and getting a refund, what a drag. 25 Dec 2025
- One of the most interesting recent privacy developments is the deployment of big two-hop IP blinding VPNs by companies like Apple and Google. These systems are designed to ensure that even those companies can’t link web requests to IP addresses. 13 Nov 2025
- Great article about how TEEs are providing much less security than folks believe they will. 29 Oct 2025 arstechnica.com
- The “age verification” and the “human identification” problem are the same problem. It upsets me to be around people who think they’re working on the first, but don’t understand they’re actually working on the second. 15 Oct 2025
- Two great bits of academic attack work this week. The new one allows Android apps to see data displayed on the screen by other apps, including Google Authenticator codes. 15 Oct 2025 pixnapping.com
- This is amazing research by Nadia Heninger and her co-authors Wenyi Morty Zhang, Annie Dai, Keegan Ryan, Dave Levin and Aaron Schulman. TL;DR a huge number of satellite links over our heads are totally unencrypted. 14 Oct 2025 satcom.sysnet.ucsd.edu
- Germany has agreed to stop ChatControl for now, due to huge amounts of public pressure. Good job! The bad news is that it could come back as soon as December, and the German government has interpreted the feedback as a need to “moderate” the proposal. 10 Oct 2025
- From yesterday: the UK is demanding another backdoor in Apple’s encryption. 2 Oct 2025 ft.com
- “Industrial-scale pig butchering scams” god please let’s just skip this century and move on to the next. 27 Sept 2025
- This battle will keep playing out over and over again until they achieve something that their own citizens have made it clear they don’t want. 6 Jul 2025 techradar.com
- A good story in WIRED giving updates on the situation with Dr. Xiaofeng Wang. 2 Apr 2025 wired.com
- The Indiana University chapter of the American Association of University Professors have written a letter demanding Dr. Xiaofeng Wang’s reinstatement. 31 Mar 2025
Link verified 20 Sept 2026. Recent items update automatically from the channel.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.