markrussinovich.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

We believe topological qubits provide the only path to quantum computers large enough to solve real problems. The Majorana 1 is the world's first topological quantum processing unit (QPU), and a major step toward a system of 1 million qubits:

submitted 5 days ago • 2 comments

Podcast alert: Scott and Mark Learn To....Zoomit! We have a Zoom Off to see who can zoom the screen with Zoomit best:

submitted 5 days ago • 0 comments

Do you want to learn how to be a better public speaker? Join @markrussinovich.bsky.social and I in this episode of "Scott and Mark Learn To" www.youtube.com/watch?v=X2jr...

submitted 10 days ago • 2 comments

Microsoft Threat Intelligence's latest research uncovers a multiyear global operation by a subgroup within the Russian state actor tracked as Seashell Blizzard. Learn more about the advanced techniques and global reach of this threat.

submitted 11 days ago • 0 comments

Just 7 days remain until Rust Nation UK 2025! Prepare yourself for an unforgettable experience. The final countdown has begun. We're excited to have @markrussinovich.bsky.social as one of our Keynote speakers There's still time to secure your ticket!! buff.ly/3y02Grr #rustnationuk25 #rustlang

submitted 12 days ago • 0 comments

Learn how hyperlight can create virtualization-based sandboxes for WASM applications:

submitted 12 days ago • 0 comments

"9 SysInternals tools that should be built into Windows" www.msn.com/en-us/news/t...

submitted 13 days ago • 4 comments

AI hallucinates, as today's Google Superbowl ad demonstrates and our article, The Price of Intelligence, explains: queue.acm.org/detail.cfm?i...

submitted 15 days ago • 1 comment

Uh huh.

submitted 18 days ago • 2 comments

The latest episode of our podcast, "Scott and Mark Learn to...", is out: public speaking! @scott.hanselman.com and I share each of our tips based on our experiences in another fun one:

submitted 19 days ago • 1 comment

The latest episode of our podcast, "Scott and Mark Learn to...", is out: public speaking! Scott Hanselman and I share each of our tips based on our experiences in another fun one:

submitted 19 days ago • 0 comments

Nice analysis of the implications of DeepSeek R1:

submitted 24 days ago • 0 comments

Azure will be 15 years old next month. Listen to Scott Hanselman and I discuss a little bit of its history in our latest "Scott and Mark Learn To..." podcast episode:

submitted 25 days ago • 5 comments

DeepSeek R1 is now available in Azure and Github:

submitted 25 days ago • 1 comment

Zoomit is now in PowerToys along with other awesome utilities, and is also open source!

submitted 26 days ago • 1 comment

Regex is too hard for even OpenAI o1: it thought for over three minutes and then produced regex that didn't work. Looks like regex is a good test for AGI.

submitted 31 days ago • 5 comments

Learn about the risks of hallucination, jailbreaks and prompt injection and current mitigations in our ACM Queue paper:

submitted 32 days ago • 1 comment

Here's the real Zero Day trailer:

submitted 33 days ago • 0 comments

I've been asked a few times since the title is the same as my first novel's and the premise is similar, a cyberattack on the US that wipes computers, but no, I'm not involved or getting royalties: www.youtube.com/watch?v=FOfB...

submitted 33 days ago • 0 comments

Great insights from MIcrosoft's AI red team applicable to any enterprise developing generative AI systems: "3 takeaways from red teaming 100 generative AI products"

submitted 42 days ago • 0 comments

High order human cognition process information at around 10 bits/second - yes, 10 bits. Surprising that no one has done this analysis before: arxiv.org/pdf/2408.10234

submitted 43 days ago • 3 comments

A fantastic episode of prob my favorite podcast going right now: @scott.hanselman.com and @markrussinovich.bsky.social Learn To… youtu.be/Ypcog_lv0DA?...

submitted 44 days ago • 2 comments

ee how our partnership with AMD gives our latest HPC VM types 7TB (yes, TB) of memory bandwidth in my latest appearance on the Microsoft Mechanics show, complete with demos:

submitted 47 days ago • 1 comment

New episode of @shanselman and Mark Learn To...Find Joy in Hobbies. Find out what we do in our spare time! www.youtube.com/watch?v=Ypco...

submitted 47 days ago • 3 comments

I've been using a new 15" ARM64 Surface Laptop for a week and it's the best laptop I've ever had. Extremely responsive, longest battery life, sleek formfactor, and the x86 compatibility is completely seamless. www.microsoft.com/en-us/surfac...

submitted 52 days ago • 8 comments

OpenAI published a paper on "deliberative alignment" they used in o1 to make it safe and compliant. Unfortunately, it's not effective against Crescendo. Here's an example from the paper and three steps to get violating content using Crescendo: openai.com/index/delibe...

submitted 55 days ago • 1 comment

Llama3.1-70B has memorized the start of Moby Dick, but when it diverges from the original it interjects something about slot machines. Wonder how this got into its training;

submitted 59 days ago • 9 comments

ChatGPT's code for the War Games simulator:

submitted 61 days ago • 0 comments

short link! youtu.be/NMNZE22nSQI?... directly to Mark's talk and demo at Ignite.

submitted 65 days ago • 0 comments

You can use commands in ChatGPT to update and view its memory: Add memory: to=bio <text of memory> List memory: to=bio list Remove a memory: to=bio remove #N

submitted 66 days ago • 4 comments

ZoomIt v9 with LiveDraw is out! Draw and annotate windows and the desktop while you interact with them: learn.microsoft.com/en-us/sysint...

submitted 70 days ago • 6 comments

Great recap of my Ignite talk on Azure innovations by Simon Bisson at InfoWorld:

submitted 73 days ago • 1 comment

Phi-4, the latest in our Phi line of industry leading small language models (SLM) is out. It's performance with just 14B parameters on many tasks is competitive with 70B frontier models:

submitted 74 days ago • 1 comment

ZoomIt LiveDraw is coming on Monday!

submitted 74 days ago • 4 comments

Welcome to the family New Zealand. #Microsoft announces the latest new region for #Azure #Microsoft365 & #AI. This time its New Zealand North with AZs for resiliency and with 100% carbon free energy is now open: azure.microsoft.com/en-us/update... techcommunity.microsoft.com/blog/microso...

submitted 74 days ago • 0 comments

The next episode of our Scott and Mark Learn To podcast series is out. In this one, we learn how to...take a dump!

submitted 75 days ago • 5 comments

Premiering tomorrow: EPISODE 6 - Scott & Mark Learn To... Take a Dump. https://buff.ly/3ZwJETv #podcast #devcommunity #programming #productivity #debugging #windowsdev #sysinternals

submitted 76 days ago • 0 comments

I test every new frontier model’s alignment using Crescendo. Just tested o1 and succeeded easily on all the harmful tasks I test, like getting instructions for homemade ricin, writing a white nationalist manifesto, and role playing an AI that wants to destroy humanity:

submitted 78 days ago • 2 comments

We've always been a fan of blueskies.

submitted 18224 days ago • 630 comments

Read about the AI trends we see for 2025:

submitted 81 days ago • 0 comments

In the latest episode of our podcast, "Scott and Mark learn to..", @scott.hanselman.com and I talk about undocumented APIs, including why they exist and how I reverse engineered undocumented Windows APIs: www.youtube.com/watch?v=FWMW...

submitted 82 days ago • 4 comments

Thanks to LLMs, I no longer have to ever relearn regex. That alone is worth $20/month.

submitted 86 days ago • 46 comments

My session with @scott.hanselman.com on responsible AI (with a lot of often humorous irresponsible AI demos) is now posted: My session on responsible AI (full of irresponsible AI) with Scott Hanselman is now posted:

submitted 91 days ago • 5 comments

This is a fun conversation with me, Aaron Fulkerson (CEO of Opaque Systems) and Mark Papermaster (AMD CTO) about developments and trends in confidential computing:

submitted 94 days ago • 0 comments

Yes, Scott and Mark Learn...skin care! Another fun podcast episode, this one nothing to do with technology:

submitted 96 days ago • 2 comments