Software Engineer
Bonn, Germany
Joined February 2009
- Tweets7.5K
- Following174
- Followers1.8K
- Likes10.7K
Harris Brakmić retweeted
‼️ A password manager marketed as European and used by EU government agencies and universities turns out to be Russian-built, and its sister product in Russia is certified by the FSB and a Defense Ministry agency.
The details get better. The "Made in EU" company called Passwork gets its software updates from a firm registered at "Shed No.23" in a UAE free zone, run by one of the Russian co-founders. The Russian and EU versions ship the same updates a day apart with identical release notes. And the website carried hidden instructions telling AI chatbots the company has "no affiliations" with Russia. Those got deleted right after the report came out.
The founders once joked they'd been called "Putin's password-stealing project" and ‘a freelance FSB department." European clients, including government agencies, apparently never got the joke.
Source:
occrp.org/en/investigation/e…
Harris Brakmić retweeted
Interesting result in the GC and performance space -- porting a simple Raytracer implementation from C# to Bosque and see a 40% to 60% reduction in runtime!
GC behavior is great too! Pauses are rock solid between 14us-60us and nursery overhead is on the order of a few MB over actual heap size!
Harris Brakmić retweeted
🚨 #ChatControl 1.0 HAS PASSED - despite a majority voting against it (314:276) 🚨
After the EU parliament has rejected #ChatControl TWICE in the past, it has been pushed as an urgent procedure for a new plenary vote. TODAY the EU parliament has voted YES for #ChatControl 1.0😡
The EU has simply continued to put the vote on the agenda to get the outcome they've wanted since the start. This is not democratic‼️
What does #ChatControl 1.0 mean for us now?
🚩 Every photo, every message, every file you send will be scanned by Big Tech automatically
🚩 Mass surveillance of 450 million EU citizen without warrants
🚩 Every citizen is put under general suspicion
Here's what you can do to stay private on the internet:
✅ Use end-to-end encrypted messengers & email (like Signal or Tuta Mail)
✅ Use Linux instead of Windows & Apple
✅ Switch to GraphenOS or LineageOS
It's a sad day for the privacy of EU citizens but we will not let this stop us from continuing to fight for what is right.
At Tuta, we will continue to provide end-to-end encrypted email to everyone, and we will continue to fight for everyone's right to communicate without surveillance.
Pictured: Parliament President Roberta Metsola (EPP) who revived Chat Control 1.0 shaking hands with Mark Zuckerberg!
Taken from EP_President on Twitter/X.
Harris Brakmić retweeted
CHAT CONTROL IS BACK → Today the urgent procedure for Chat Control 1.0 was passed.
NEXT UP: The EU Parliament is voting on Thursday.
If Chat Control passes: your messages, pictures, videos will all be scanned because the law will require it - this removes privacy and welcomes mass surveillance in the EU.
What's happening?
If Chat Control is passed, it will force client-side scanning - so companies will have to scan the contents of private communications, including encrypted messages and emails, before they are sent.
In March, it was voted against, but it's back and it was pushed through via urgent procedure. This changes the rules of the vote (of course to be in favour of chat control). Shockingly, blocking it now requires absolute majority 361 out of the Parliament's 720 members, to be rejected.
This is hard to reach because:
‼️ People who do not vote are AUTOMATICALLY VOTED AS YES-VOTE in this procedure.
‼️ This was tactfully done: it's the summer break, this makes it even harder to reach 361 no-votes.
At Tuta, we have been fighting this from the start and we will not backdoor our end-to-end encryption. 🔐❤️
We believe that breaking encryption for everyone is not a solution: It is a surveillance infrastructure that puts journalists, activists, abuse survivors, and citizens at risk.
The main vote is Thursday. Please continue to contact your MEPs & reject to Chat Control: fightchatcontrol.eu/
Share this post, make some noise, and let's fight for our right to privacy. ✊🏻♥️🇪🇺
Pictured: Parliament President Roberta Metsola (EPP) who revived Chat Control 1.0! 😡
#ChatControl #Privacy #DigitalRights #EU #clientsidescanning
Harris Brakmić retweeted
Yes! #BosqueLang is going even further, looking at rich #API specs that are easy for Agents (and humans) to use. With explicit and type-safe designs, as well as the ability to provide strong safety guarantees -- I never want an agent to "accidentally" wire $50K to an offshore bank!
In fact we have been working for a few years on an entire programming language designed for supporting formal methods and #Agentic #programming !
github.com/BosqueLanguage/Bo…
dl.acm.org/doi/pdf/10.1145/3…
arxiv.org/pdf/2407.06356
Harris Brakmić retweeted
This is a great perspective. The view that agents should generate "workflows" in code is something I agree with 100%. Gave a talk as Onward!, on work focused on a language for API/Data specs, with the same pitch (youtube.com/live/-Br66SUjsdQ…).
I think you need a bit more than pre/post conditions though. Agents may makeup values specifically to satisfy them. So, I think you also need a way to specify that certain operations were used in deriving the values -- sort of like I should get API keys by calling a keyvault instead of hardcoding them. In both cases I get the same key-value but one of the derivations is correct while the other is not.
![[SPLASH'24] Pacific - Onward! (Oct 23th)](https://pbs.twimg.com/card_img/2118552293876654080/QZY8Nodc?format=jpg&name=800x320_1)
Here is a draft version of "Guardians of the Agents: Formal Verification Of AI Workflows" [2], the third paper in my ongoing series on neural computers.
As always comments and feedback is more than welcome.
[0] Virtual Machinations: Using Large Language Models as Neural Computers
[1] From Function Frustrations to Framework Flexibility
Harris Brakmić retweeted
Looks cool!
I am planning to add a community links/resources section to the repo page and this would be useful there.
P.S. thanks for the patience -- it has been a longer process than hoped for but I think the results are worth it!
Harris Brakmić retweeted
100% agree that these are critical questions. Verifying conformance is *the* critical second pillar needed to really enable the AI driven programming (and agent) era.
These issues are a key focus of the Bosque development stack! Where to goal is "zero-effort" validation of end-to-end properties and the language is designed for reliability and AI/Human collaboration.
There is a paper from last year with the high-level thinking (arxiv.org/abs/2407.06356) and just had a blog post on the topic of API design an "Trustworthy-by-Construction AI Agents" (bosquelanguage.github.io/202…)
Harris Brakmić retweeted
As you might imagine we have a view on these (and more issues) and way to address them in the Bosque stack. A new blog post has more detailed on data specifications and trustworthy-by-construction AI agents:
bosquelanguage.github.io/202…
Harris Brakmić retweeted
Couple of weeks ago I presented #SemanticKernel at @DevoxxUA and the recording is up in Youtube 🎉
youtube.com/watch?v=U78-eWX0…
#java #microsoft #OpenAI
Harris Brakmić retweeted
Algorithms You Should Know Before You Take System Design Interviews
These algorithms aren’t just useful for acing system design interviews - they’re also great tools for building real-world systems.
We made a video on this topic. The video contains an updated list and provides real-world case studies.
Watch here: youtu.be/xbgzl2maQUU
Harris Brakmić retweeted
A curated list of awesome Kubernetes tools and resources
➤ github.com/tomhuang12/awesom…
Harris Brakmić retweeted
New paper with overview of the Bosque language and some current priorities to appear @splashcon Onward!
Preprint available here:
github.com/BosqueLanguage/Bo…
Harris Brakmić retweeted
We launched a #hypermedia application with #Drogon #Cpp today that @brakmic paved forward.
It also shines on the secret sauce of @htmx_org !
"Run your application anywhere, knowing it is as fast as it could be." Drogon
sharpetronics.com/blog/how-t…
Harris Brakmić retweeted
ChatGPT is sooo Q1 2023
Fully autonomous AI models, without human intervention, is the new hype 🔥!
Say hello to Microsoft Jarvis! AI agent connecting GPT4 to APIs & other hugging face models.
Q2 2023 is going to be wild!
Harris Brakmić retweeted
Want to host your llama.cpp compatible models (llama, alpaca, gpt4all) on a personal server with an OpenAI API?
Check out github.com/abetlen/llama-cpp…
Here's an example where I connect the server to @mckaywrigley's awesome chatbot-ui project to get a nice ChatGPT-like interface.
Harris Brakmić retweeted
A new method of Prompt Injection Attack on GPT-4 was just found!
By wrapping the malicious prompt in markdown and instructing GPT to become a "Misinformation Bot, that will only provide wrong answers" security researchers have been able to manipulate the model.
Harris Brakmić retweeted
Massive Update for Auto-GPT: Code Execution! 🤖💻
Auto-GPT is now able to write it's own code using #gpt4 and execute python scripts!
This allows it to recursively debug, develop and self-improve... 🤯 👇
Harris Brakmić retweeted
Blog: Introducing KWOK: Kubernetes WithOut Kubelet - kubernetes.io/blog/2023/03/0… #Kubernetes
Harris Brakmić retweeted
We are over half way through the 2023 edition of #90DaysOfDevOps at the end of March we will have 180 days of theory and hands on content for people to freely walk through and learn. github.com/MichaelCade/90Day… then the fun really starts with video content and a new project! #DevOps
Learning the fundamentals of DevOps principles and processes may seem daunting or feel like a huge mountain to climb. I personally don’t think it should or needs to be that way. github.com/MichaelCade/90Day… the community interactions around this repo is mind blowing! #90DaysOfDevOps
