Ever thought about how a cyberattack without malware can still wreak havoc? The recent Europe refrigeration hacks show how critical it is to secure connected OT/IoT. A must-watch for anyone in cybersecurity or infrastructure 🔒 #OT#DefrostGate#ServiceNow
https://t.co/c6bo2R7Zcv
Netflix’s “Untold” documentary series will revisit the Giants’ historic upset of the undefeated Patriots in Super Bowl XLII.
“Untold: One Giant Upset” features firsthand accounts from Eli Manning, Tom Brady, Michael Strahan and Tom Coughlin on one of the biggest upsets in NFL history.
They did it! SpaceX has just caught their Starship rocket for the first time ever, marking the first recovery of a conventional orbital rocket upper stage in history!
This is one of the greatest engineering achievements in history. Congrats to the entire @SpaceX team! 🚀
I have a major result to share.
I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.
This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.
Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.
I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.
I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.
It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.
These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.
I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at https://t.co/YOtzND0Jqk.
There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.
The White House said it has nominated Michael LoSapio to be the director of the US Department of Homeland Security's Cybersecurity and Infrastructure Security Agency https://t.co/jVtL4HhyNB
Osi Umenyiora says he discovered Tom Brady's pre snap cadence and knew when he was going to snap the ball ahead of Super Bowl XLII, and says it allowed the Giants defensive line to dominate the game with 5 sacks and 20 hurries
“But as I was watching the tape, because you have two weeks leading up to the game, I noticed that Tom Brady would come up to the line of scrimmage. And this is real information I’m giving you here, worth millions of dollars. I’m giving it to you right now. And there’s people recording this to pass it on to their players.”
“But what I noticed was he would come up to the line of scrimmage and he would say ‘hut,’ which is telling the center to snap him the ball. So what he was doing was he would come up to the line of scrimmage, and before he said that, he was pointing at somebody, maybe somebody on the opposing side. And if he didn’t point at that person, that means they weren’t going to snap the ball. But if he did point at that person, the ball was coming as soon as he said ‘hut.’”
“So after I watched this, I studied this for literally a week. I went back, I watched all the games, and he was doing the exact same thing, but nobody had figured it out. So I saw that, and I went and gathered the whole defense, and I was like, ‘Listen, this is what this man is doing.’ If you know when the ball is being snapped as a defensive line, you are at a tremendous advantage over the offensive lineman because they’re relying on you not knowing.”
“Because if you don’t know, that means you’re going to be late. But if you do know, you’re going to beat them almost every time. And so as soon as I figured this out, I gathered the whole defensive line. I was like, ‘Listen, this is what Tom Brady’s doing.’”
“Now we all started to watch this together, and we saw this was 100%. It wasn’t even like 80. This was 100%. So now we knew going into that game that, as a defensive line, we were going to be able to overwhelm them offensively simply because of that one thing that I had pointed out.”
“And we went into the game and we sacked him like five times, hit him like 20 times. They couldn’t understand how we were getting off the ball and getting to him that fast. And it was simply because, for two weeks prior, I just sat there and I had watched what this one person was doing, and that gave us the mental advantage going into the game. We knew we were going to win that game even though nobody else did, because we knew something that they didn’t know, and nobody else knew that we knew.”
CISA 2030 just dropped, and OT is written right into it. 🛡️ Defend · 🧱 Harden · ⚡ Prevail, with 60-day, 180-day, and 1-year milestones. My take: you can't evict what you can't see. Can you identify which OT assets exist? 👇 #CISA2030#OTSecurity#ConnectingTheDOTS
They say everything is bigger in Texas, but Skattebo don’t give a damn!
Sometimes you gotta bully the bully, and he started in the first quarter and didn’t stop until the Giants walked it off with the game-winning run in overtime!
People got ran over, tackles got broken, and celebrations were on repeat. My man had folks in compromising positions all afternoon! 😂😂
@thepivot sports on @youtube.
#ThePivotSports #ThePivot #Skattebo #Giants
https://t.co/gUelSpfuuc
The vulnerability tsunami is coming in just a few months and there won’t be enough time to patch critical infrastructure. Network isolation described below (air-gaps and data diodes!) needed now!
So in fact we actually can get robots, flying cars, superintelligence, fast internet on planes, space travel, space colonies, orbital data centers, unlimited energy, likely cures for nearly every disease, and more all in our life times. Pretty great.
7/ Do you know every path into your controllers, including the ones your vendors installed?
I break down one OT security story every two weeks. Follow to keep connecting the dots. #OTSecurity
6/ 3 moves for CISOs:
1. Get internet-exposed OT to zero, vendor modems included
2. Route remote access through a gateway with MFA (NIST SP 1800-45 shows how)
3. Rehearse manual ops and prove you can recover