Matthew Green
@matthewdgreen.bsky.social
📤 18643
📥 407
📝 2246
I teach cryptography at Johns Hopkins.
https://blog.cryptographyengineering.com
reposted by
Matthew Green
Toby Murray
2 days ago
Matt call this an investigation of a $10 question. That undersells the importance of this encryption to frontier labs. It’s one of their primary defences against model distillation attacks, which represent major threats to their competitiveness.
add a skeleton here at some point
0
25
5
My kid was joking the other day about visiting conspiracy websites, like the ones that end in .gov.
2 days ago
0
54
6
Last week I discovered that ChatGPT and Claude will send you their “encrypted raw reasoning” and of course I immediately wasted a weekend trying to do something bad with it. What I got for my trouble was this blog post:
blog.cryptographyengineering.com/2026/05/29/f...
loading . . .
Fooling around with encrypted reasoning blobs
This is a quick post I wanted to write about a “hobby project” I spent a weekend on. It has little to do with real cryptography, and mostly doesn’t expose a particularly exciting …
https://blog.cryptographyengineering.com/2026/05/29/fooling-around-with-encrypted-reasoning-blobs/
2 days ago
3
64
23
So I sort of found an issue with the OpenAI and Anthropic APIs, but they disagree. I think that means I can blog about it?
3 days ago
5
46
0
One of the things I used to like about Matthew Yglesias is that he criticized evidence-free windbags like Thomas Friedman. Now he’s an evidence-free windbag and he thinks he’s killing it.
6 days ago
2
32
2
reposted by
Matthew Green
Matthew Gracie
7 days ago
Yeah, this. I also miss peak Infosec Twitter. I know this isn't it, but neither is current Twitter.
add a skeleton here at some point
2
52
5
Last Thursday, the Texas AG Ken Paxton filed suit against WhatsApp, alleging some fairly serious (but highly nebulous) “privacy violations” in WhatsApp. I was going to write a blog post about this, but then I remembered I already had — when these allegations first surfaced.
7 days ago
2
29
11
The snark from Codex is making me uneasy.
8 days ago
9
62
4
My son is doing an internship where (against all my attempts to persuade him into a different career) he’s looking for security vulnerabilities in Chinese-made medical devices, and oh boy is he finding them. I feel like this is unhealthy validation for an 18 year old’s first job.
9 days ago
8
221
25
reposted by
Matthew Green
Riana
9 days ago
NCMEC's full 2025 CyberTipline report is out. 1.5MM reports "with a GAI [generative AI] nexus," but removing Amazon's useless 1.1MM reports, >182K reports of possession or (attempted) generation of GAI CSAM. That's the number to report: 182K, not 1.5 million.
www.missingkids.org/gethelpnow/c...
loading . . .
CyberTipline Data
https://www.missingkids.org/gethelpnow/cybertipline/cybertiplinedata
1
7
7
Every single one of them could have been a great name for our kids.
9 days ago
6
27
1
Spent ten minutes trying to get Anthropic to accept my CC number to buy credits. Turns out you had to click “billing address is not the same as shipping address” and re-enter your whole address, only then can you click Continue. The quality of everything around Claude is an anti-ad for Claude.
9 days ago
15
130
18
How not to write a reassuring email 101. We meet Tuesdays 6-10pm.
11 days ago
2
46
4
There’s been some reporting that Meta contributed an unfathomable sum to promote age verification laws globally. This is broadly true, but the actual situation is a bit more complex. Figured it was worth an update.
13 days ago
1
58
22
A good primer on the new Bitlocker exploit.
solcyber.com/bitlocker-in...
loading . . .
BitLocker in crisis? The "YellowKey" zero-day in plain English - SolCyber
Nightmare Eclipse hates Microsoft, loves dropping 0-days.
https://solcyber.com/bitlocker-in-crisis-the-yellowkey-zero-day-in-plain-english/
14 days ago
2
57
24
I’m going to confess that I love using coding LLMs. It’s basically taken software development from “cool ideas I don’t have time for” to “let’s throw together a prototype.” But I also don’t think anyone’s reviewing the production code they’re producing, at least not thoroughly.
15 days ago
34
232
20
When you ask AI to improve a picture.
16 days ago
1
24
1
The kids wanted to make a website for our dog Darwin. Turns out Darwin is a tough domain name, so we ended up here.
dogw.in
loading . . .
Dogw.in
Dogw.in, an original 8-bit mini dachshund game.
https://dogw.in/
17 days ago
1
29
2
I don’t quite know what to tell you about that.
20 days ago
1
8
1
reposted by
Matthew Green
JHU Computer Science
23 days ago
JHU’s
@matthewdgreen.bsky.social
talks to the
@nytimes.com
about the potential risks of
@microsoft.com
Copilot’s health records tool: “There is a pot of gold of high-value data that is in one location that people can get.”
loading . . .
A.I. Chatbots Want Your Health Records. Tread Carefully.
https://www.nytimes.com/2026/03/12/technology/personaltech/microsoft-copilot-health-ai-chatbots.html
1
7
7
reposted by
Matthew Green
Anna Leigh 🏳️‍⚧️
21 days ago
Remote attestation is a heavily underdiscussed threat to computing freedom. People often mistakenly dismiss it with "I will run my own fork (of the OS or browser) / Magisk", not understanding that with hardware attestation, you're literally unable to 1/
add a skeleton here at some point
8
352
173
I think it’s weird that even Apple sends me SMS text messages rather than iMessage, from their automated systems.
24 days ago
3
28
1
Me talking to ChatGPT, after learning that the Beale ciphers are probably fake.
25 days ago
1
12
1
The latest iOS beta has support for end-to-end encryption between Apple and Android devices via RCS text messaging.
25 days ago
3
84
18
My experience being an AI-native software developer for four weeks (AKA a project manager for Codex and Claude Code) makes me laugh at Bryan Acton’s shtick: “get your hands dirty alongside their teams.” Nobody’s getting their hands dirty.
26 days ago
4
16
0
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.
26 days ago
8
102
30
I’ve been working on this classical cipher cryptanalysis tool using a mixture of Claude Code and Codex, just to get a sense of how capable these tools are on big (advanced) code bases. So far it solves homophonic, substitution, and a few polyalphabetic test cases.
github.com/matthewdgree...
loading . . .
GitHub - matthewdgreen/decipher: An AI-enabled application for cracking ciphers
An AI-enabled application for cracking ciphers. Contribute to matthewdgreen/decipher development by creating an account on GitHub.
https://github.com/matthewdgreen/decipher
26 days ago
3
30
4
I had never looked at the Kryptos ciphers before, but they turn out to be Vigenere with an alternative shift alphabet: KRYPTOSABCDE… The coding tools want to build crib searchers and I want to do frequency analysis and annealing to see if that can be recovered statistically.
about 1 month ago
1
12
0
🍩
loading . . .
about 1 month ago
3
39
0
I swear, every day I learn about a new type of Claude Code usage limit: daily, weekly, monthly. More types of usage limit than my dogs have chew toys.
about 1 month ago
3
55
7
We’re getting Zodiac Z340 decrypts!
add a skeleton here at some point
about 1 month ago
2
36
7
I just taught the boy how to get out of vim.
about 1 month ago
3
75
7
I think it’s fascinating that we’re entering a time where the value of a lot of proprietary software tooling goes to zero, and AI companies should be the first people to realize this, but instead they’re doubling down on gobbling up proprietary software companies like Cursor.
about 1 month ago
5
27
5
Last Friday I took the 18y/o out against his will to find the most interesting dining experience in Baltimore. We found it at Lithuanian Hall, where we had Ceppilinai. Also I’m a member now.
about 1 month ago
3
45
4
As a second Claude Code hobby project I’ve been trying to get it to crack classical crypto. Turns out this requires assembling good benchmarks from a lot of public material. But at a certain point Claude Code decided that continuing on this project represented a violation of their “Usage Policy”.
about 1 month ago
5
78
13
Weird that a bunch of call options expire on 4/17 and suddenly the whole Hormuz situation looks better.
about 1 month ago
1
35
5
reposted by
Matthew Green
Fredrik Dahlgren
about 1 month ago
Two weeks ago, Google published a paper proving in zero-knowledge that they had an efficient implementation of Shor's algorithm. Today, Trail of Bits can prove that we have an even better implementation which beats Google's on all metrics! 🫢
blog.trailofbits.com/2026/04/17/w...
1
83
37
reposted by
Matthew Green
hikikomorphism
about 2 months ago
I have a long-running Claude session just for talking about Iran war news/economic shocks from the blockade. I shared the latest Trump tweet and it told me that it can no longer keep role-playing this fictional scenario because it's becoming concerning and detached from reality. Love this timeline.
9
235
37
Everything on crypto Twitter/X is quantum computing hype. It’s amazing how obviously artificial this all is.
about 2 months ago
4
29
3
reposted by
Matthew Green
Filippo Valsorda
about 2 months ago
Alright, it's official! đź’°
@matthewdgreen.bsky.social
and I bet on what will break first, ML-KEM-768 or X25519. The loser donates to a 501(c)(3) picked by the winner. If you have an opinion on quantum computers or lattices, you can join with a side bet. Just submit a PR!
github.com/FiloSottile/...
add a skeleton here at some point
7
119
31
reposted by
Matthew Green
Zack Whittaker
about 2 months ago
New: France said it plans to move its government computers currently running Windows to the open-source software Linux to further reduce its reliance on U.S. tech. Comes at a time of growing instability and unpredictability on the part of the Trump administration and weaponization of sanctions, etc.
loading . . .
France to ditch Windows for Linux to reduce reliance on US tech | TechCrunch
France's move to ditch Windows for Linux is its latest effort to reduce its reliance on American tech giants.
https://techcrunch.com/2026/04/10/france-to-ditch-windows-for-linux-to-reduce-reliance-on-us-tech/
6
234
130
reposted by
Matthew Green
Nigel Smart
about 2 months ago
Betting on the Quantum Apocalypse...
www.theregister.com/2026/04/09/c...
loading . . .
https://www.theregister.com/2026/04/09/cryptograhpers_quantum_bet/
1
9
4
reposted by
Matthew Green
Zack Whittaker
about 2 months ago
New, by me at TechCrunch: The developer of the widely popular Wireguard VPN says he is also unable to ship software updates to Windows users after Microsoft locked his account, marking the second high-profile app developer (VeraCrypt) in the past few weeks to face this issue.
loading . . .
WireGuard VPN developer can't ship software updates after Microsoft locks account | TechCrunch
The popular open source VPN maker is the second high-profile developer to say Microsoft locked his account without notifying him and are blocking their ability to send software updates to users.
https://techcrunch.com/2026/04/08/wireguard-vpn-developer-cant-ship-software-updates-after-microsoft-locks-account/
4
109
77
We should be able to slow down AI takeover by a few years just by telling the model to find every bug in ffmpeg.
about 2 months ago
5
84
9
reposted by
Matthew Green
Paul Frazee
about 2 months ago
I feel like we're not addressing the most concerning news from Mythos
27
560
91
reposted by
Matthew Green
Juliet Shen
about 2 months ago
this is honestly THE best write up of how CSAM detection and perceptual handing works. the visual aids are very helpful in understanding how content is transformed and how detection methods work
mahmoud-salem.net/the-invisibl...
loading . . .
How Do You Find an Illegal Image Without Looking at It?
61.8 million files of suspected child abuse were reported in 2025 alone. This is how machines detect them at internet scale — without any human ever seeing the content.
https://mahmoud-salem.net/the-invisible-shield
8
212
91
I think this is a good precautionary analysis but I’d bet huge amounts of money against a relevant quantum computer by 2029 or even 2035.
add a skeleton here at some point
about 2 months ago
12
161
49
Does anyone over here know anything about which stablecoins Iran might be using to demand tolls for Hormuz transit?
www.bloomberg.com/news/article...
loading . . .
Secret Codes and Yuan Fees Get Ships Through Iran’s Hormuz Tollbooth
Vessels wanting to transit the strategic waterway need to be from friendly countries, and some have to pay fees in Chinese currency or crypto before being escorted through the strait.
https://www.bloomberg.com/news/articles/2026-04-01/strait-of-hormuz-ships-paying-iran-yuan-and-crypto-tolls-for-safe-passage?embedded-checkout=true
about 2 months ago
5
19
6
Well, at least blockchains are getting exciting again.
about 2 months ago
3
12
0
reposted by
Matthew Green
Natanael, Tech janitor
about 2 months ago
gizmodo.com/group-pushin...
loading . . .
Group Pushing Age Verification Requirements for AI Turns Out to Be Sneakily Backed by OpenAI
It gave the leader of a nonprofit involved with it
https://gizmodo.com/group-pushing-age-verification-requirements-for-ai-turns-out-to-be-sneakily-backed-by-openai-2000741069
1
19
8
Load more
feeds!
log in