This significantly changes the context of the “iOS Vuln dev hacked!” story. It is a clear national security issue, with exploits sold to the U.S. being resold to Russia. Both Russia and the U.S. (e.g. FBI) would have legitimate reasons to investigate the suspected leaker. https://t.co/6FCyEyWxtl

— thaddeus e. grugq (@thegrugq) October 23, 2025

If these two stories are related then it is extremely likely that the iPhone hack was part of an investigation into someone believed to be selling exploits to Russia. Both the FBI and Russia would have compelling reasons to check out the accused leaker.

— thaddeus e. grugq (@thegrugq) October 23, 2025

"Some people will try to convince you that prompt injection attacks can be solved using more AI to detect the attacks. This does not work 100% reliably, which means it’s not a useful security defense at all.

The only solution that’s credible is to run coding agents in a…

— Dino A. Dai Zovi (@dinodaizovi) October 23, 2025

If you're excited to see the WhatsApp bug thrown @thezdi - free to watch my talk from @reconmtl 2025 on 4 remote bugs I discovered last year!

While they're not 0-click RCE - there are some remote corruption and funny logic bugs in there.https://t.co/N78H5QeNNZ

— Luke (datalocaltmp) (@datalocaltmp) October 23, 2025

"To secure AI Applications, we must first solve identity". Some great ideas from @jalkove and industry colleagues on where we need to be spending deep thinking time as Agents evolve in the Enterprise. https://t.co/Xs0GnxDHck

— Heather Adkins - Ꜻ - Spes consilium non est (@argvee) October 22, 2025

AI agents with "human approval" protections can be bypassed with argument injection. We achieved RCE across three platforms by exploiting pre-approved commands like git, ripgrep, and go test. 🧵 pic.twitter.com/30JobvQ2zW

— Trail of Bits (@trailofbits) October 22, 2025

Credential Guard was supposed to end credential dumping. It didn't.@bytewreck just dropped a new blog post detailing techniques for extracting credentials on fully patched Windows 11 & Server 2025 with modern protections enabled.

Read for more ⤵️ https://t.co/mYPHg1mTKj

— SpecterOps (@SpecterOps) October 23, 2025

So basically what I take away from this is that there's a way to get around Credential Guard by using Remote Credential Guard to give you a NTLMv1 response challenge that you can then crack.

I admit I too was under the impression that NTMLv1 doesn't work with Cred Guard. https://t.co/D7JsePoA4M pic.twitter.com/9nG6vxKcpK

— Brian in Pittsburgh (@arekfurt) October 24, 2025

The blog links a free NetNTLMv1 cracking site (https://t.co/qgzIVr4roR) whose operator appears tied to Chengdu Mistiny Ltd., a company in Sichuan, China. I don't know if you really want to let them crack your hash values. https://t.co/x5lqT6xOJH pic.twitter.com/h0qDoyTWkn

— Florian Roth ⚡️ (@cyb3rops) October 24, 2025

pic.twitter.com/cE3pSZklzA

— Trend Zero Day Initiative (@thezdi) October 23, 2025

European security services are investigating whether a former Brussels-based Politico reporter was a spy for China, according to officials from three separate intelligence agencies. The reporter, whose identity is known to Euractiv, no longer works for Politico. It is not clear…

— Bill Bishop (@niubi) October 23, 2025

Proofpoint threat researchers have designed an open-source tool—named PDF Object Hashing—to track and detect the unique characteristics of PDFs used by threat actors... similar to a digital fingerprint. 🫆

We use this tool internally to help track multiple threat actors with… pic.twitter.com/jV0VFeGdD0

— Threat Insight (@threatinsight) October 23, 2025

#ESETresearch discovered a new wave of the well-known 🇰🇵 Lazarus campaign Operation DreamJob, now targeting the drone industry. @pkalnai @alexis_rapin https://t.co/lR9FTFnCCN 1/9

— ESET Research (@ESETresearch) October 23, 2025

Perplexity Comet’s AI assistant can take screenshots of websites and analyze them for users.

However, Comet will also follow instructions hidden on a webpage that it screenshots. pic.twitter.com/m1Dvle9veX

— Brave (@brave) October 21, 2025

call this cluster messy panda. https://t.co/44Eo47m47F pic.twitter.com/9keaFtj4Nv

— J⩜⃝mie Williams (@jamieantisocial) October 23, 2025

From friend - current US in late Qing malaise

Flooded with opioids
Closed to foreign trade
Navy rotten to the core
Government in paralysis
Misuse of military funding
Governors ignoring central orders
Throwing birthday parties for the emperor
Fixing up imperial gardens pic.twitter.com/uBI0zJe74d

— Lawrence Zhang 張樂翔 (@HistorianZhang) October 24, 2025

The German company that built the forklift used in the Louvre robbery uses a photo of the heist to advertise itself: “When things need to be done quickly.”

😂😂😂 pic.twitter.com/Ucd17nUm7X

— BowTiedMara (@BowTiedMara) October 23, 2025

🚨 NEW PAPER 🚨: “From Chaos to Capability: Building the U.S. Market for Offensive Cyber”⁰by myself and @SergeyBratus⁰👉 https://t.co/fDOWmEY6Nr
Should the U.S. outsource its cyberattacks? We talked to 30 experts across gov, VC, and industry to find out. pic.twitter.com/0yKGx5YkfQ

— Winnona 💾 (@__winn) October 20, 2025

$1,024,750 - 73 unique bugs - a week of amazing research on display. #Pwn2Own Ireland had it all. Success. Failure. Intrigue. You name it. Congratulations to the Master of Pwn winners @SummoningTeam! Their outstanding work earned them $187,500 and 22 point. See you in Tokyo for… pic.twitter.com/Vxd5b0yJ55

— Trend Zero Day Initiative (@thezdi) October 24, 2025

Now I don’t know a lot about OnlyFans, but I do know those 42 people aren’t the ones generating the revenue https://t.co/S8uAwURZoZ

— Dr. Wesley McGrew (@McGrewSecurity) October 24, 2025