전사 🔗
WEBVTT
00:00:10.400 --> 00:00:13.839
Here is your briefing for Tuesday, November 17th, 2026.
00:00:15.440 --> 00:00:24.160
A trusted website script on CNN, the BBC, and Zoom spent Monday skimming cards, swapping crypto addresses, and aiming exploits at older iPhones.
00:00:27.920 --> 00:00:31.440
A trusted script on CNN and Zoom turned into a skimmer.
00:00:31.679 --> 00:00:41.039
An attacker exploited a flaw in Optimizely's website experimentation service and planted a loader inside the custom code that customer sites pull straight from Optimizely's own servers.
00:00:41.199 --> 00:00:50.719
Report URI founded in scripts belonging to at least 178 organizations, including the BBC, Salesforce, GitLab, IKEA, HSBC, and Equifax.
00:00:50.879 --> 00:00:58.000
Optimizely says the flaw is fixed and the code is gone, but it hasn't said how the attacker got in or whether they still have access.
00:00:58.159 --> 00:01:01.039
And one company script got reinfected after it was cleaned.
00:01:02.079 --> 00:01:12.960
The payload swapped crypto wallet addresses on the page and even in your clipboard grabbed card numbers typed into site forms and on older iPhones loaded two known exploit kits, Karuna and Darksword.
00:01:13.200 --> 00:01:18.880
Researchers haven't confirmed any phone actually got taken over, and the attacker's wallets collected about$7,000.
00:01:19.520 --> 00:01:22.400
Card fields inside a payment provider's own frame were safe.
00:01:22.560 --> 00:01:23.120
Do this today.
00:01:23.200 --> 00:01:27.760
If your site uses optimizely, fetch every snippet you serve and search it for the indicators in the show notes.
00:01:27.920 --> 00:01:36.160
Rotate every user and token that can write to your account, keep third-party tags off checkout pages, and update every iPhone and iPad to the latest release.
00:01:39.439 --> 00:01:42.800
NCentral finally has a fix that's supposed to hold last week.
00:01:43.040 --> 00:01:49.120
Attackers beat Enable's second hotfix and dropped crypto miners on NCentral servers that had already patched.
00:01:49.280 --> 00:01:59.840
Now Enable has shipped Hotfix 3, which it calls a critical security update, covering 17 more flaws across logins agent and probe traffic, file handling, and user management.
00:02:00.640 --> 00:02:07.280
The big change is that the legacy API setting, the door the root exploit needed, has been ripped out entirely.
00:02:07.520 --> 00:02:15.840
Hosted customers get the update automatically, but on-premises servers have to upgrade, and Enable warns that agents and probes left behind can be impersonated.
00:02:16.080 --> 00:02:19.439
Remote control and the Intune integration stay blocked for now.
00:02:19.680 --> 00:02:20.400
Do this today.
00:02:20.560 --> 00:02:23.680
Install Hotfix 3 on every on-premises and central server.
00:02:23.759 --> 00:02:31.599
Push the latest agent and probe to every managed device, and keep hunting for miners and unknown accounts, because a patch won't evict anyone already inside.
00:02:34.960 --> 00:02:37.439
Your patch notes are now on attacker's to-do list.
00:02:37.680 --> 00:02:41.840
Google's threat intelligence team says attackers have moved past chatting with AI models.
00:02:41.920 --> 00:02:52.960
They're now running open source, multi-agent attack frameworks that handle whole stages of an intrusion on their own at machine speed, and both criminal crews and suspected China-linked spies are experimenting with them.
00:02:53.680 --> 00:02:55.840
The scarier part is patch diffing.
00:02:56.000 --> 00:03:06.400
Attackers feed AI pipelines, the old code, the new code, the release notes, and any public research and get a working exploit far faster than before, which Google saw with paper cut flaws in September.
00:03:06.560 --> 00:03:16.080
Google also flags stolen AI keys being resold cheap through underground proxy services and a Chinese phishing kit with a built-in chatbot that customizes lures on request.
00:03:16.319 --> 00:03:25.280
Do this today, shrink your patch window for anything internet facing to days, not weeks, starting with every fix that shipped this month, and set spending alerts on every AI provider account.
00:03:28.879 --> 00:03:32.000
Supermarket fridges are hackable and there's no patch.
00:03:32.240 --> 00:03:40.319
Researchers at Midnight Blue found six flaws in Dan Foss's flagship store refrigeration controller that let a remote attacker take one over without logging in.
00:03:40.480 --> 00:03:42.719
Five of them sit in code that runs as root.
00:03:42.879 --> 00:03:47.200
The researcher found all six in under an hour, and the newest firmware doesn't fix them.
00:03:48.080 --> 00:03:59.200
These controllers were already attacked in France starting in October, and a second wave across Europe last week forced defrost cycles and destroyed stock, though nobody has tied these six new flaws to those attacks.
00:03:59.360 --> 00:04:03.120
There's no secure boot either, so planted code could survive a factor reset.
00:04:03.280 --> 00:04:05.920
Belgium, Ireland, and Germany are all saying the same thing.
00:04:06.159 --> 00:04:08.240
Get them off the internet, do this today.
00:04:08.480 --> 00:04:22.000
If you run or support grocery, coal storage, or restaurant sites, pull every Dan Foss controller's management page off the public internet, put remote access behind a virtual private network, and check set points, defrost schedules, and user accounts for changes you didn't make.
00:04:25.430 --> 00:04:29.269
Log for Shell is still opening doors, this time to a code signing server.
00:04:29.509 --> 00:04:35.430
Kaspersky detailed a month-long breach at a Russian software developer by Bio Team, a group assessed to be Ukrainian.
00:04:35.600 --> 00:04:48.560
They came in through an exposed remote login service on a Linux server, then took over an old VMware vCenter that was still vulnerable to Log4Shell, got root, and planted an hourly backdoor on the hypervisor dressed up as a PostgreSQL cache job.
00:04:49.360 --> 00:04:58.480
From there, they reach the build pipeline and the server that signs the company's software releases, which Kaspersky reads as groundwork for a supply chain attack on the developer's customers.
00:04:58.639 --> 00:05:09.040
They mostly use built-in tools and legitimate accounts, and after getting kicked out, they phish staff with emails that pass sender checks using a key stolen from the company's own mail server.
00:05:09.280 --> 00:05:09.920
Do this today.
00:05:10.000 --> 00:05:21.040
Confirm no vCenter you run is still vulnerable to log4 shell, check hypervisor scheduled task folders for scripts you didn't put there, and move code signing keys into hardware protection with an alert on every signing request.
00:05:24.160 --> 00:05:33.120
Hijacked marketing scripts, a hotfix that needed a hotfix AI turning patches into exploits, fridges with no fix, and a five-year-old bug guarding a signing server.
00:05:33.280 --> 00:05:37.199
The code you load, sign, and forget about is exactly the code attackers want.
00:05:37.280 --> 00:05:39.360
And one quick note for Screen Connect shops.
00:05:39.519 --> 00:05:45.759
ConnectWise says attacks are still landing wherever those third party extensions stay enabled, so update and leave them off.
00:05:48.959 --> 00:05:50.000
That's your brief.
00:05:50.160 --> 00:05:52.879
Stay sharp, patch your systems, and we'll see you tomorrow.