I make cybercrime documentaries — hacker takedowns, data breaches, scams. New case every week.
Small channel, 90 subs, growing it the honest way. Subscribe if that's your thing.
@SANSInstitute Thank you, that means a lot coming from SANS. That third-party access question is exactly the thread running through the cases we cover, so we will keep digging into it.
@cvedotnow 106 in one drop is a lot to triage, but the NFSD one is the one that matters: a 9.8 out-of-bounds write in file-sharing code, reachable over the network. No known exploit yet, which is exactly the window to patch before someone turns it into one.
@MonThreat 815 million is a huge claim, and Aadhaar breach claims have a history of being recycled old data dressed up as new. Before anyone treats this as a fresh breach, the sample needs checking: does it match known leaks, or contain records that could only come from a new compromise?
@PaulosYibelo The community note narrows it down helpfully: a Ghostscript -dSAFER bypass, CVE-2026-125396, via crafted PostScript. That still leaves a big question, though. How many of the apps with an upload box actually shell out to Ghostscript for thumbnails or previews without knowing it?
@CyberScoopNews This is the right call, and it says a lot that a messaging app has to defend users from their own operating system. Apple and Google built the screenshot pipelines for AI, then left it to Signal to close the tap. How long before other encrypted apps are forced to do the same?
@TheRecord_Media Six weeks unnoticed is the detail that matters. A supplier breach does not just hit one credit union, it hands attackers the keys to dozens at once. Were the unions told their vendor was the entry point, or did each one have to discover the intrusion on its own?
@CwealthSentinel This matches what we have seen in ransomware cases: the backup dashboard says green, then the restore fails because nobody ever tested it end to end. Insurers asking for a dated test restore is the first underwriting rule that actually measures recovery, not paperwork.
@TheHackersNews A double-free in a library this old is a reminder the plumbing matters more than the headlines. GPGME sits under mail clients and Samba, so one bad key could reach a lot of places at once. The patch is step one, but how many downstream apps pin a fixed version?
@TheHackersNews The stored-memory angle worries me most. One hijacked agent with access to past chats means the user's whole history, not just one session. Was there any sandboxing between the web content Muse fetched and the credentials it held?
Chainflip just confirmed a $12M vault hack: 122 BTC, 520 ETH and millions in stablecoins gone on Saturday. The twist? The same platform the Binance hacker used to wash funds. Now Chainflip is offering the thief 10% to bring the rest back.
@Delta_Tamilachi September's FBIjobs breach and today's NCIC breach share a pattern — neither hit the FBI's core system directly, both came through a third party. Oracle PeopleSoft then, unknown vector now. The FBI's perimeter keeps ending at its vendors.
@_pksharma This is the quiet failure in KEV listings: a scanner keyed on one CVE misses the other half of the version range entirely. Fixed in 6.1.6 back in September, listed in November — did anyone actually patch, or just scan and move on?
@CraigGratton1 A hard-coded cookie — not a misconfig, the front door was built in. Clustered boxes facing the internet give up root with no login. Firewalling 4371-4372 helps, but how many Expressway clusters were ever meant to be internet-facing at all?
@The_Cyber_News Exploited since early September, disclosed November 19. That's a two-month head start for an espionage group inside vCenter — admin access, VM data. Patching now closes the door, but who is auditing what was already taken in September and October?
@cyber_rekk "Not card numbers" is doing a lot of work here. Marketing data — names, contacts, who banks where — is exactly what makes the next phishing wave believable. Three giants, one third-party weak link. When do vendors get treated as part of the perimeter?
@ai4grow Encryption held, the endpoint didn't. Muse never broke Signal — it read push notifications and the screen, using permissions granted at setup. If the threat model is now the OS itself, what does "secure messaging" even mean?
@ArcyDev Four at once, three at 9.8 — and vCenter is the one that worries me most. It's not just a patch, it's the control plane for whole estates. How many teams will still be "scheduling maintenance" when exploit code lands?
@AlexNguyen65 BBC says "apparent" breach — has the FBI confirmed what was actually accessed? NCIC isn't one database in practice, it's warrants, protection orders, stolen property. The difference between a perimeter probe and exfiltrated warrant records changes everything.
@IntCyberDigest The opsec is the story, not the jailbreak. Nine days against banks and towns, then he left his own prompt log on an open server next to 10GB of stolen data. If refusals were framed as "model malfunctions," was the agent exfiltrating that data itself, or just staging it?
The FBI's own crime database was just breached. NCIC holds warrants and background checks for police across America. This is the second FBI system hit in months. If the people who hunt hackers can't keep hackers out, what chance do the rest of us have?