# SECITHUB

The hub for cybersecurity Industry

- Site: https://secithub.com/

## Autonomous AI Cyberattack | How exploit_agent Hit French Websites With No Human in the Loop

- Link: https://secithub.com/autonomous-ai-cyberattack-france/
- Published: 2026-11-18T17:33:39+02:00
- Author: SECITHUB

Reading Time:   5 min

The first **autonomous AI cyberattack** may already be behind us. On November 16,
2026, Google’s Threat Intelligence Group (GTIG) reported that Russian-speaking hackers
used an AI system called “exploit_agent” to find and exploit software flaws with
no human in the loop. The targets were French websites in the media, entertainment
and education sectors. Google says it is the first time it has seen an attack chain
run end to end by AI agents.

For years, security experts warned that AI would one day hack on its own. That day
appears to have arrived quietly, on an exposed server, against mid-sized websites.
The bigger news is not who was hit. It is how little skill the operators needed 
and how fast the tool turned patches into working exploits.

## **What Google Found**

The discovery came from a mistake by the attackers.

### **An exposed staging server**

GTIG researchers found a server that the operators had left open. It held the exploit_agent
framework, stolen data from victims and configuration files. The victim data showed
a clear focus on French sites.

### **A chain of AI agents**

The framework links several AI agents together. It works with both frontier models
and models hosted on the attackers’ own hardware. Each agent handles one stage of
the attack, and the chain runs without a person approving each step.

### **Patch in, exploit out**

The system watches popular web software for new security patches. It compares the
code before and after each fix, works out what the change protects against and writes
an exploit for the old version. Then it runs that exploit against real targets on
its own.

### **Tools for the next stage**

Researchers also found tools for deeper access after the first break-in. One tool
removes web shells planted by rival hackers. That lets the operators push competitors
off a server they have taken over.

    ```wp-block-verse
    The patch you release today is the exploit someone's AI writes tonight.
    ```

## **Who Is Behind the Autonomous AI Cyberattack?**

Attribution is limited, and it is worth reading carefully.

### **Russian-language clues**

The configuration files contain comments written in Russian. One tells the user 
to enter API keys for “auto-exploitation” in the file rather than in a chat. That
points to Russian-speaking operators. It does not prove links to the Russian state
or to any known group.

### **What “Russian-speaking” really means**

Russian is a working language across much of the cybercrime world, from Russia and
Belarus to Ukraine, Kazakhstan and beyond. Language clues help narrow the field,
but they rarely settle who ordered an attack. Criminal crews, hacktivists and state
units all write code in Russian. Careful analysts treat this kind of evidence as
a lead, not a verdict.

### **A tense backdrop**

The report landed the same day the EU sanctioned 30 people and two entities over
Russian information manipulation and cyber operations, including the pro-Russian
group NoName057(16). France has faced repeated Russian-linked hacking and disinformation
campaigns. No one has tied the sanctions to this AI operation, but the timing adds
to concern in Paris.

### **Not the only one**

Google also found a China-linked group building a similar autonomous framework, 
most likely for espionage. The same report describes stolen AI service credentials
and tampered settings in AI coding tools. Autonomous offense is becoming a trend,
not a one-off.

## **Why This Changes the Patching Race**

Defenders have long relied on a gap between a patch and a working exploit. That 
gap is closing fast.

### **Every patch is a roadmap**

Security researchers have always diffed patches to find the flaw behind them. It
took skill and days of work. An AI agent can now do the same in hours and repeat
it for every update across many products at once.

### **Low skill, high output**

Google warns that the barrier to entry is low and attackers will very likely keep
using such frameworks. A small crew with API keys can now run an operation that 
once needed a team of exploit developers. Mid-sized sites that never drew skilled
attackers are now in range, because an AI agent does not care whether a target is
famous.

### **Speed leaves traces**

There is one piece of good news. Google says these frameworks trade stealth for 
speed. They leave noisy forensic traces, such as bursts of exploit attempts right
after a patch. Teams that watch for that pattern can catch them.

### **Same tools, both sides**

Defenders are using AI too. In the same week, researchers used **_[GPT-6 Cyber to find dozens of Apple flaws](https://secithub.com/gpt-6-cyber-apple-vulnerabilities/)_**
before criminals did. The race now favors whoever automates first.

## **What Website Owners and Security Teams Should Do**

Treat the time between a vendor patch and your deployment as live exposure. Every
day an update waits in a queue is a day an AI agent can use it against you.

 1. **Patch web software within days, not weeks.** Content management systems, plugins
    and web frameworks are prime targets for patch-diffing agents.
 2. **Turn on automatic updates** where the risk of breakage is low, especially for
    security-only releases.
 3. **Use a web application firewall** with virtual patching to cover the gap before
    you can update.
 4. **Watch for post-patch spikes.** Alert on surges of unusual requests to recently
    patched endpoints.
 5. **Hunt for web shells.** Scan servers for unknown files and scripts, including 
    ones another attacker may have left.
 6. **Protect your AI keys.** Rotate API keys, restrict them by IP and monitor usage.
    Our article on **_[exposure-based triage](https://secithub.com/ox-radar-response-center/)_**
    shows how to decide what to fix first.

    ```wp-block-verse
    When attackers automate, your patch window becomes your attack surface.
    ```

The autonomous AI cyberattack on French websites was not the largest or most damaging
attack of the year. Its importance lies in what it proves. Machines can now find
a flaw, build the exploit and use it without a human hacker at the keyboard.

That shifts the advantage to speed. Organizations that patch fast, monitor closely
and automate their own defenses will keep pace. Those that patch on a monthly cycle
will find AI agents waiting for them.

Want to see how quickly AI-found flaws reach critical software?

**_[Read our deep-dive: Java4Shell CVE-2026-114938.](https://secithub.com/java4shell-cve-2026-114938/)_**

    ```wp-block-verse
    Close your patch window before an AI agent walks through it.
    ```

![SECITHUB FAQ banner for the first autonomous AI cyberattack on French websites](
https://secithub.com/wp-content/uploads/2025/12/SECITHUB-FAQ-1.jpg)

**What is the first autonomous AI cyberattack?**

Google reports that Russian-speaking hackers used an AI framework called exploit_agent
to find and exploit flaws in French websites without human involvement.

**What is exploit_agent?**

A chain of AI agents that compares software before and after security patches, writes
exploits for the fixed flaws and runs them against targets automatically.

**Who was targeted?**

French websites in the media, entertainment and education sectors, according to 
victim data on the attackers’ server.

**Who is behind the attack?**

The operators appear to speak Russian, based on comments in the code. Google has
not linked them to a known group or to the Russian state.

**Can autonomous AI attacks be detected?**

Yes. Google says these frameworks trade stealth for speed and leave noisy traces,
such as bursts of exploit attempts right after patches.

**How can website owners protect themselves?**

Patch web software within days, enable automatic security updates, use a WAF with
virtual patching, monitor post-patch traffic and hunt for web shells.

### **Sources**

_[Google says Russian-speaking hackers ran the first fully automated AI cyberattack – CyberScoop](https://cyberscoop.com/google-fully-automated-ai-cyberattack/)_

_[Google says attackers used AI to turn patches into exploits with no human – The Next Web](https://thenextweb.com/news/google-gtig-ai-exploits-patches-no-human)_

The post [Autonomous AI Cyberattack | How exploit_agent Hit French Websites With No Human in the Loop](https://secithub.com/autonomous-ai-cyberattack-france/)
appeared first on [SECITHUB](https://secithub.com).

## Deepfake Zoom Scam | How North Korea Clones CEOs to Hack Executives

- Link: https://secithub.com/deepfake-zoom-scam-north-korea/
- Published: 2026-11-18T17:26:40+02:00
- Author: SECITHUB

Reading Time:   5 min

A North Korean **deepfake Zoom scam** is spreading faster than ever. New research
from Arctic Wolf Labs shows the BlueNoroff group has targeted 677 people in 58 countries
by cloning the faces and voices of real executives and using them in fake video 
calls. About 84% of the targets are founders or senior leaders in tech, Web3 and
venture capital. Days earlier, the FBI issued its own alert on the same scheme, 
which it calls “FaceStealer.”

The scam does not break any software. It breaks trust in the one thing many people
still treat as proof: a live face on a video call. If your executives speak at conferences
or appear on YouTube, attackers already have the raw material to impersonate them.

## **How the Deepfake Zoom Scam Works**

The campaign follows a clear, repeatable playbook.

### **Step 1: Harvest public footage**

The attackers collect interviews, podcasts and conference talks of founders and 
CEOs. Public speakers give them hours of clean video and audio to work with.

### **Step 2: Clone the executive**

Arctic Wolf says the group uses commercial AI tools to clone faces and voices. According
to the researchers, the results can no longer be told apart from real webcam video
by simple visual inspection.

### **Step 3: Book a fake meeting**

Victims receive an invite to what looks like a Zoom or Microsoft Teams call with
a known partner or investor. In earlier cases, the first contact came through a 
compromised Telegram account belonging to a real executive. The meeting link leads
to a look-alike site.

### **Step 4: Stage a “technical problem”**

During the call, the fake executive reports audio or video trouble. The victim is
told to install an “update,” a plugin or a fix, or to paste a command into a terminal.
That step installs the malware.

### **Step 5: Turn victims into bait**

Here is the most dangerous twist. The attackers quietly record the victim’s own 
webcam during the call. They then use that footage to impersonate the victim to 
their contacts. Each new victim becomes the lure for the next one.

    ```wp-block-verse
    Every victim's face becomes the next victim's bait.
    ```

## **Who Is Behind It and Why It Is Growing**

BlueNoroff is a financial theft unit linked to North Korea’s Lazarus Group. Its 
main goal is stealing cryptocurrency to fund the regime.

### **The numbers keep climbing**

In April 2026, the campaign targeted just over 400 people in 20 countries. Today
it reaches 677 targets in 58 countries. Arctic Wolf tracked 98 malicious domains,
most of them newly registered. Analysts also flagged 349 videos as likely manipulated.

### **The FBI is watching**

On November 12, the FBI published a FLASH alert on “FaceStealer,” describing fake
Zoom, Teams and Google Meet calls run by North Korean operators. The alert pulls
together months of vendor research into one law enforcement account of the full 
attack chain.

### **AI multiplies, it does not reinvent**

Arctic Wolf makes an important point. AI acts as a production multiplier for the
group, but it has not changed how the intrusion works. The victim still has to run
something. That means the defense still works if people follow it.

### **A huge prize**

US officials say North Korea stole more than $2 billion in cryptocurrency in 2025.
A single compromised founder or fund partner can open the door to wallets, treasury
keys and investor funds. That makes executives in crypto and venture capital some
of the most valuable targets in the world.

## **Why Deepfake Calls Work So Well**

These attacks exploit habits that most professionals never question.

### **Video feels like proof**

Many security policies tell staff to “verify by video” when an email looks odd. 
Deepfakes turn that advice into a weakness. A familiar face now proves nothing on
its own.

### **Tech problems are normal**

Everyone has joined a call with broken audio. A request to install a quick fix sounds
routine, especially from a busy executive. Attackers rely on that politeness and
urgency, much like the **_[help desk scams used by Scattered Spider](https://secithub.com/gta-6-hacker-guilty-scattered-spider/)_**.

### **Public profiles feed the machine**

Founders and investors are expected to be visible. They speak on panels, record 
podcasts and post videos to build their brands. Every clip gives attackers more 
material for a convincing clone. Executives do not need to disappear online, but
security teams should know which leaders have the largest public footprint and brief
them first.

### **Detection tools lag behind**

Deepfake detection products exist, but quality improves faster than detectors can
keep up. Relying on software to spot a fake face is a losing race. Process controls,
such as call-back rules and install bans, work no matter how good the fake looks.

## **How to Protect Your Team From Deepfake Calls**

Most of these steps cost nothing but discipline.

 1. **Never install anything during a call.** No legitimate Zoom or Teams meeting requires
    a plugin, update or terminal command from a participant.
 2. **Use only official apps and links.** Join meetings from the real Zoom, Teams or
    Meet app, and check the domain before you click.
 3. **Verify through a second channel.** Confirm unusual meetings or requests with 
    a call to a known number or a message on a separate platform.
 4. **Agree on code words.** Executives and finance teams can use private phrases to
    confirm identity on sensitive calls.
 5. **Lock down endpoints.** Block unsigned software and script execution on executive
    laptops, and monitor macOS as closely as Windows.
 6. **Train leadership first.** Executives are the main targets, so brief them directly.
    Our guide to **_[the human attack surface](https://secithub.com/human-attack-surface-2025/)_**
    shows how to start.

    ```wp-block-verse
    If a call asks you to install something, end the call.
    ```

The deepfake Zoom scam shows how quickly AI has erased the line between real and
fake on screen. A face and a voice are no longer enough to prove who you are talking
to.

The good news is that the final step still depends on the victim. Organizations 
that set one firm rule, never install or run anything because someone on a call 
asked, can stop this campaign cold.

Want to see how AI is changing the speed of both attack and defense?

**_[Read our analysis: GPT-6 Cyber Finds 35 Apple Flaws.](https://secithub.com/gpt-6-cyber-apple-vulnerabilities/)_**

    ```wp-block-verse
    Trust the process, not the face on the screen.
    ```

![SECITHUB FAQ banner for the North Korean deepfake Zoom scam targeting executives](
https://secithub.com/wp-content/uploads/2025/12/SECITHUB-FAQ-1.jpg)

**What is the deepfake Zoom scam?**

A North Korean campaign that clones real executives’ faces and voices and uses them
in fake Zoom or Teams calls to trick victims into installing malware.

**Who is behind the deepfake Zoom scam?**

BlueNoroff, a financially motivated unit linked to North Korea’s Lazarus Group. 
The FBI calls the scheme FaceStealer.

**Who is being targeted?**

Arctic Wolf counted 677 targets in 58 countries. About 84% are founders or senior
executives in tech, Web3 and venture capital.

**How does the malware get installed?**

During the call, the fake executive claims a technical problem and asks the victim
to install an update or plugin or run a command.

**Why is the scam spreading so fast?**

Attackers record victims’ webcams during calls and reuse the footage to impersonate
them to new contacts.

**How can I protect my organization?**

Never install software during a call, join only through official apps, verify requests
through a second channel and block unsigned software on executive devices.

### **Sources**

_[FaceStealer: Deepfake Video Meetings as a Self-Propagating Intrusion Chain – Cloud Security Alliance](https://labs.cloudsecurityalliance.org/wp-content/uploads/2026/11/CSA_research_note_facestealer_deepfake_meeting_intrusion_20261114-csa-styled.pdf)_

_[North Korean hackers deepfake execs in Zoom call to spread Mac malware – BleepingComputer](https://www.bleepingcomputer.com/news/security/north-korean-hackers-deepfake-execs-in-zoom-call-to-spread-mac-malware/)_

The post [Deepfake Zoom Scam | How North Korea Clones CEOs to Hack Executives](https://secithub.com/deepfake-zoom-scam-north-korea/)
appeared first on [SECITHUB](https://secithub.com).

## GTA 6 Hacker Pleads Guilty | What Lapsus$ and Scattered Spider Teach About Help Desk Attacks

- Link: https://secithub.com/gta-6-hacker-guilty-scattered-spider/
- Published: 2026-11-18T17:21:04+02:00
- Author: SECITHUB

Reading Time:   5 min

The **GTA 6 hacker** case has reached its final chapter just as the game itself 
hits shelves. On November 16, 2026, Arion Kurtaj, now 21, pleaded guilty at Woolwich
Crown Court in London to the 2022 hack of Rockstar Games and a string of other attacks
he carried out as a teenager. Meanwhile, TechCrunch reports that Tyler Buchanan,
a 24-year-old from Dundee, has received a six-year US sentence for a phishing scheme
that stole millions in cryptocurrency.

Both men belonged to the same loose world of young hackers behind Lapsus$, 0ktapus
and later Scattered Spider. Their crimes did not rely on advanced malware. They 
relied on phone calls, fake login pages and help desks that trusted the wrong caller.
That is the part every security team should study.

## **The GTA 6 Hacker Pleads Guilty**

Kurtaj’s case is one of the strangest in recent cybercrime history.

### **The Rockstar breach**

In September 2022, Kurtaj broke into Rockstar Games and stole internal data, including
about 90 clips of early Grand Theft Auto VI footage. He leaked the clips and tried
to blackmail the company. Rockstar refused to negotiate and later said the hack 
cost it about $5 million to recover from.

### **Hacking from a hotel room**

At the time, Kurtaj was on bail for earlier hacks and staying in a hotel under police
protection. He had no laptop. He used an Amazon Fire TV stick, a phone and the hotel
TV to break into Rockstar’s systems.

### **A long road to a guilty plea**

In 2023, a jury found that Kurtaj carried out the attacks, and a court ordered him
held in a secure hospital. In 2026, doctors found him fit to stand trial, and he
moved to prison ahead of a retrial. His plea this week covers attacks on Rockstar
Games, BT and EE, Revolut and Uber. He will be sentenced at a later hearing.

    ```wp-block-verse
    A Fire TV stick and a hotel TV were enough to breach one of the world's biggest game studios.
    ```

## **The Scattered Spider Sentence**

Buchanan’s case shows the same playbook at a larger, more profitable scale.

### **Fake help desks, real losses**

Buchanan and his co-conspirators sent waves of text messages and made calls that
posed as company IT teams. Employees entered their credentials on fake login pages,
and the group used them to break into corporate networks. Prosecutors say the crew
then stole at least $8 million in cryptocurrency from victims.

### **From arrest to prison**

Spanish police arrested Buchanan at Palma airport in June 2024. He was sent to the
United States in April 2025 and pleaded guilty in April 2026 to wire fraud conspiracy
and aggravated identity theft. He is the last of five defendants sentenced in the
case. A co-defendant, Noah Urban, received ten years in 2025.

### **A group running out of members**

Law enforcement has steadily taken apart the network. In July 2026, a UK court jailed
two other Scattered Spider members over the 2024 attack on Transport for London.
Authorities say the arrests have disrupted the group’s core operations, though copycats
still use its name.

## **What the GTA 6 Hacker Case Teaches Defenders**

Researchers nicknamed these crews “advanced persistent teenagers.” They hit Microsoft,
NVIDIA, Uber, MGM Resorts and dozens of others without nation-state tools.

### **People are the entry point**

These groups rarely needed a software exploit. They called help desks, sent SMS 
lures and spammed users with MFA prompts until someone approved one. The weakest
link was a busy employee, not a server. Our analysis of **_[the human attack surface](https://secithub.com/human-attack-surface-2025/)_**
explains why this keeps working.

### **Collaboration tools are the new crown jewels**

Kurtaj reached Rockstar through its internal messaging platform. Once inside Slack,
Teams or similar tools, attackers can read secrets, find admin contacts and impersonate
staff. Many companies still protect these tools less carefully than email.

### **Age is not a defense signal**

Teenagers with free time, social skills and online forums caused major damage across
these campaigns. Defenders should not dismiss “kids” as low-risk. Their methods 
scale because they target processes that every company shares.

### **AI makes the old tricks stronger**

The calls that fooled help desks in 2022 relied on confidence and a good script.
Today, attackers can add cloned voices, fluent messages in any language and fake
profiles built in minutes. That makes caller ID and “sounds like the CEO” useless
as proof of identity. Verification has to rely on something the attacker cannot 
copy.

## **How to Stop Help Desk and Phishing Attacks**

The fixes are well known, and most are cheaper than a single breach.

 1. **Verify every reset request.** Require call-backs to known numbers, manager approval
    or in-person checks before help desks reset passwords or MFA.
 2. **Move to phishing-resistant MFA.** Use passkeys or FIDO2 security keys instead
    of SMS codes and push prompts.
 3. **Stop MFA fatigue.** Turn on number matching and limit repeated prompts.
 4. **Protect phone numbers.** Ask carriers for SIM-swap locks on executive and admin
    lines.
 5. **Lock down chat platforms.** Enforce SSO, monitor new sessions and keep secrets
    out of messages.
 6. **Train with real scenarios.** Run fake help desk calls and SMS lures, not just
    email tests. Our **_[identity security guide](https://secithub.com/active-directory-security-hardening-identity/)_**
    covers the controls behind these steps.

    ```wp-block-verse
    Your help desk is a security control. Treat every call like it could be the breach.
    ```

The GTA 6 hacker case and the Scattered Spider sentence close a chapter in one of
the most damaging cybercrime waves of the decade. Courts are catching up, and the
core members now face prison.

But the techniques live on. New crews, and now AI voice tools, make help desk fraud
and phishing easier than ever. The real lesson is not about one teenager in a hotel
room. It is about how easily trusted processes can be turned against the companies
that run them.

Want to see how another group is using leak sites to pressure big brands?

**_[Read our analysis: The Gentlemen Ransomware Claims NVIDIA.](https://secithub.com/the-gentlemen-ransomware-nvidia/)_**

    ```wp-block-verse
    Harden the human layer before the next caller asks for a password reset.
    ```

![SECITHUB FAQ banner for the GTA 6 hacker guilty plea, the Scattered Spider sentence
and help desk social engineering defense](https://secithub.com/wp-content/uploads/
2025/12/SECITHUB-FAQ-1.jpg)

**Who is the GTA 6 hacker?**

Arion Kurtaj, a Lapsus$ member from Oxford who hacked Rockstar Games in 2022 and
leaked early Grand Theft Auto VI footage. He pleaded guilty on November 16, 2026.

**How did the GTA 6 hacker breach Rockstar?**

While on bail and staying in a hotel under police protection, he used an Amazon 
Fire TV stick, a phone and the hotel TV to access Rockstar’s internal systems.

**What did Kurtaj plead guilty to?**

Attacks on Rockstar Games, BT and EE, Revolut and Uber, carried out when he was 
a teenager. He will be sentenced at a later hearing.

**Who is Tyler Buchanan?**

A 24-year-old from Dundee, Scotland, linked to Scattered Spider. He was sentenced
to six years in the US for a phishing scheme that stole millions in cryptocurrency.

**What is Scattered Spider?**

A loose group of mostly young English-speaking hackers known for help desk social
engineering and SMS phishing. It overlaps with Lapsus$ and 0ktapus.

**How can companies stop help desk social engineering?**

Verify reset requests through known channels, use phishing-resistant MFA, block 
MFA fatigue, protect phone numbers from SIM swaps and train staff with realistic
call scenarios.

### **Sources**

_[British hackers face prison for GTA 6 data breach and crypto thefts – TechCrunch](https://techcrunch.com/2026/11/18/british-hackers-face-prison-for-gta-6-data-breach-and-crypto-thefts/)_

_[British hacker tied to Scattered Spider pleads guilty – The Record](https://therecord.media/hacker-scattered-spider-guilty-plea)_

The post [GTA 6 Hacker Pleads Guilty | What Lapsus$ and Scattered Spider Teach About Help Desk Attacks](https://secithub.com/gta-6-hacker-guilty-scattered-spider/)
appeared first on [SECITHUB](https://secithub.com).

## OX Radar and Response Center | Why “Does This Threat Affect Us?” Is the New Security Question

- Link: https://secithub.com/ox-radar-response-center/
- Published: 2026-11-18T17:06:32+02:00
- Author: SECITHUB

Reading Time:   5 min

OX Security has launched **OX Radar** and a new Response Center, two tools built
around one question every security team asks after a big disclosure: does this threat
actually affect us? The company announced both on November 18, 2026. OX Radar watches
for new vulnerabilities, poisoned packages and actively exploited threats. The Response
Center then checks each one against a customer’s own code and environment.

The timing fits the moment. In a single week, defenders faced a critical Java runtime
bug, a poisoned Microsoft package on PyPI and dozens of AI-found Apple flaws. Most
teams cannot read every advisory, let alone check every system by hand. Tools that
answer “are we exposed?” in minutes rather than days are becoming a core part of
the security stack.

## **What OX Radar and the Response Center Do**

The two tools work as a pair. One watches the outside world. The other looks inward
at the customer’s environment.

### **OX Radar watches the threat feed**

OX Radar scans CVE databases, known-exploited vulnerability catalogs, security advisories
and vendor research about every ten minutes. For customers, it focuses on compromised
or malicious dependencies, exploited vulnerabilities and critical CVEs. A separate
public feed tracks new npm and PyPI packages, AI skills and Model Context Protocol(
MCP) servers.

### **The Response Center checks your exposure**

When a threat appears, the Response Center maps it to the customer’s assets. It 
finds which repositories, libraries and services are affected and adds context and
fix guidance. Teams can start an investigation from a CVE ID, a GitHub Security 
Advisory ID or an advisory URL.

### **Signals it combines**

 * CISA and ENISA known-exploited vulnerability catalogs
 * EPSS scores, which predict the chance of exploitation
 * VEX statements on whether a product is truly exploitable
 * AI-generated exploit checks and runtime visibility from the OX platform

### **From finding to fix**

Teams can rank risk by severity, active exploitation, code usage and reachability.
They can then open tickets, pick safe versions and create fix pull requests from
the same place. The tool also keeps a record of each response, which helps with 
audits and post-incident reviews.

    ```wp-block-verse
    The question is no longer "is this threat dangerous?" It is "is it dangerous to us?"
    ```

## **Why “Does It Affect Us?” Is Now the Key Question**

OX CEO Neatsun Ziv framed the launch around relevance rather than raw danger. That
shift reflects a real problem across the industry.

### **Too many alerts, too little time**

Tens of thousands of new CVEs appear every year. AI-assisted research is adding 
even more, faster. Security teams spend hours checking advisories, asset lists and
scanner output just to learn whether a new threat applies to them. Many alerts turn
out to be noise, but nobody can know without checking.

### **Severity alone misleads**

A CVSS score of 9.8 means little if the vulnerable function never runs in your code.
A medium bug can be urgent if attackers already exploit it and it sits on an internet-
facing service. Real priority depends on three things: is it exploited, is it present,
and is it reachable?

### **Supply chain attacks move in minutes**

Poisoned packages now appear and vanish quickly. This week’s **_[Microsoft Agent Framework PyPI compromise](https://secithub.com/microsoft-agent-framework-pypi-compromise/)_**
lasted about 55 minutes. A daily scan would miss it entirely. Feeds that refresh
every few minutes give teams a better chance to react before damage spreads.

## **Where OX Radar Fits in the Market**

OX is not alone in chasing this problem. Still, its approach shows where application
security is heading.

### **Part of a wider platform**

Former Check Point executives Neatsun Ziv and Lior Arzi founded OX Security in 2021,
after the SolarWinds supply chain attack. The company calls its platform an AI-native
application protection platform (AINAPP). It covers code scanning, cloud security,
AI agent security and an agentic pentester. OX Radar and the Response Center plug
into that platform for existing customers.

### **A crowded and fast-moving field**

Application security posture management (ASPM), software composition analysis and
cloud security vendors are all adding exploitability and reachability context to
their tools. The market is moving away from long lists of findings toward short,
ranked answers. For buyers, the real test is accuracy on fresh threats, speed of
the first verdict and how well the tool fits existing developer workflows.

### **Vendor claims need testing**

OX has not published detection rates, false-positive figures or independent tests
for the new tools. Buyers should run them against real recent threats, such as Java4Shell
or the PyPI compromise, and compare the results with their own manual findings.

## **How to Build “Does It Affect Us?” Into Your Process**

You can improve exposure triage with or without a new product.

 1. **Know what you run.** Keep an up-to-date SBOM for every application and container
    image.
 2. **Prioritize by exploitation.** Put KEV entries and high EPSS scores at the top
    of the queue, above raw CVSS.
 3. **Check reachability.** Confirm whether the vulnerable code path actually runs 
    before you escalate.
 4. **Watch your registries.** Monitor npm, PyPI and MCP servers for malicious or hijacked
    packages that your teams use.
 5. **Automate the fix path.** Link findings to tickets and pull requests so fixes 
    ship fast.
 6. **Measure time to answer.** Track how long it takes to say “affected” or “not affected”
    after a major disclosure. Our **_[supply chain security guide](https://secithub.com/supply-chain-security-guide-2025/)_**
    explains how to build this capability.

    ```wp-block-verse
    The fastest team is not the one that patches everything. It is the one that knows what to patch first.
    ```

OX Radar and the Response Center are a vendor answer to an industry-wide problem.
Whether they deliver will depend on real-world accuracy, not launch claims. But 
the idea behind them is sound. Exposure, not volume, should drive the security queue.

As AI floods the world with new vulnerabilities, the teams that win will be those
that can separate the threats that matter to them from the noise, and act on them
within hours.

Want to see what a fast-moving supply chain attack looks like up close?

**_[Read our deep-dive: Java4Shell CVE-2026-114938 | How an AI-Found Base64 Bug Puts Enterprise Java at Risk.](https://secithub.com/java4shell-cve-2026-114938/)_**

    ```wp-block-verse
    Know your exposure before the next disclosure lands in your inbox.
    ```

![SECITHUB FAQ banner for OX Radar, the OX Security Response Center and exposure-
based vulnerability prioritization](https://secithub.com/wp-content/uploads/2025/
12/SECITHUB-FAQ-1.jpg)

**What is OX Radar?**

OX Radar is a threat monitoring service from OX Security. It scans vulnerability
databases, exploited-vulnerability catalogs and advisories about every ten minutes
for threats that may affect customers.

**What does the OX Security Response Center do?**

It checks a new threat against a customer’s code and environment, finds affected
assets and provides context, prioritization and fix guidance.

**What signals does the Response Center use?**

CISA and ENISA known-exploited catalogs, EPSS exploit predictions, VEX exploitability
statements, AI-generated exploit checks and runtime visibility.

**Is OX Radar available to everyone?**

The Response Center and customer OX Radar are part of the OX Security platform. 
A broader OX Radar research feed is public on the OX Security website.

**Why is reachability important in vulnerability prioritization?**

A vulnerable library only matters if the affected code runs in your application.
Reachability analysis helps teams skip noise and fix real risks first.

**How can teams triage new threats faster without new tools?**

Maintain SBOMs, prioritize by KEV and EPSS, check reachability, monitor package 
registries and automate tickets and fix pull requests.

### **Sources**

_[OX Security launches OX Radar and Response Center – Cybersecurity Insiders](https://www.cybersecurity-insiders.com/ox-security-launches-ox-radar-and-response-center-to-show-companies-which-cyber-threats-actually-affect-them/)_

_[Known Exploited Vulnerabilities Catalog – CISA](https://www.cisa.gov/known-exploited-vulnerabilities-catalog)_

The post [OX Radar and Response Center | Why “Does This Threat Affect Us?” Is the New Security Question](https://secithub.com/ox-radar-response-center/)
appeared first on [SECITHUB](https://secithub.com).

## Vintage Infrastructure | Why Hacked Supermarket Fridges Are a Warning for Critical Systems in the AI Era

- Link: https://secithub.com/vintage-infrastructure-hacked-fridges/
- Published: 2026-11-18T15:28:26+02:00
- Author: SECITHUB

Reading Time:   5 min

**Vintage infrastructure** has become one of the biggest security risks of 2026,
and Europe’s hacked supermarket fridges show why. In early November, attackers raised
the temperature of fridges and freezers in dozens of stores across at least seven
countries, some to 50°C. No novel exploit was needed. According to Cisco’s chief
security and trust officer Anthony Grieco, the attackers used old, known weaknesses
in systems that should have been retired or locked down long ago.

Grieco’s warning goes beyond supermarkets. The same mix of ageing, unsupported technology
and AI-powered attackers threatens hospitals, factories, water utilities and networks.
In his view, the risk equation has changed sharply in just the last six months.

## **What Happened to Europe’s Supermarket Fridges**

The fridge attacks came in two waves and hit some of Europe’s best-known grocery
brands.

### **A first strike in France**

In October, an attack on refrigeration systems in France caused food losses and 
disruption. At the time, it looked like an isolated incident.

### **A second wave across Europe**

Overnight into November 2, stores in Belgium, Estonia, Ireland, Finland, Luxembourg,
the Netherlands and Sweden found their cooling systems running hot. An ICA store
in Sjöbo, Sweden, said its fridges and freezers had been set to 50°C. Delhaize, 
Coop and other chains pulled chilled and frozen goods and threw away stock.

### **The real cost**

Retailers threw away tons of spoiled food, with losses estimated in the millions
of euros. Some stores closed for hours to clean and reset their equipment. Smaller
shops were hit too. A butcher in Athens said someone pushed one of his refrigeration
units to 150°C. For businesses that run on thin margins, one night of tampering 
can wipe out weeks of profit.

### **Still no culprit**

Investigators have not named the attackers or confirmed a link between the two waves.
Authorities have not published technical indicators or a CVE tied to the attacks.
What is clear is how the attackers worked: they reached the controllers that manage
temperature remotely and changed the settings.

    ```wp-block-verse
    It took a few commands to a temperature controller to empty the freezers of a continent.
    ```

## **Why Vintage Infrastructure Is the Real Problem**

Grieco’s central point is simple. The danger is not new hacking techniques. It is
old systems nobody secured properly.

### **What “vintage” means**

Vintage infrastructure is hardware or software that has reached end of life. The
vendor no longer ships security updates, yet the system keeps running. Grieco compares
it to a 30-year-old car that still drives, even though the original brake pads are
no longer made.

### **Connected without being secured**

Refrigeration controllers, building systems and factory equipment often join the
network for remote monitoring. Many of them use default passwords, flat networks
and protocols built decades ago. IT teams rarely track them the way they track laptops
and servers, so these devices fall outside patch cycles and security reviews.

### **The same risk, higher stakes**

A spoiled freezer costs money. A compromised infusion pump, assembly line or water
treatment plant can cost far more. Grieco says the same pattern of old, poorly secured
systems now runs through hospitals, factories, water utilities and digital networks.

### **Why old systems stay in place**

Replacing industrial equipment is expensive and disruptive. A controller that works
can stay in service for 20 years or more. Budgets favor new projects over quiet 
upgrades, and every swap risks downtime. That is how unsupported systems pile up
until an attacker finds them.

## **How AI Makes Vintage Infrastructure More Dangerous**

Old weaknesses are not new. What has changed is how fast attackers can find and 
use them.

### **Attackers move at machine speed**

Grieco says attackers now use AI tools and agents to break into systems far faster
than humans could. AI can scan the internet for exposed devices, match them to known
flaws and automate the attack. It also makes phishing and social engineering more
convincing.

### **AI finds hidden flaws, too**

Researchers have used AI models to uncover long-hidden bugs in major platforms, 
from **_[the Java runtime](https://secithub.com/java4shell-cve-2026-114938/)_** 
to Apple’s operating systems. Old firmware that nobody has audited in years is an
even easier target.

### **Defenders can use AI as well**

Grieco argues the same technology helps defenders work at a new pace. Cisco says
its teams used AI tools to scan 1.8 billion lines of code in eight weeks, work it
estimates would take engineers about eight years. The company also works with partners
using frontier models from OpenAI and Anthropic to find and fix flaws first. These
are vendor figures, but the direction is clear.

## **What Organizations Should Do Now**

Grieco’s advice is practical, and most of it costs less than a single bad night 
in a supermarket.

 1. **Find every connected device.** Inventory controllers, sensors, cameras and building
    systems, not just computers. You cannot protect what you do not know exists.
 2. **Retire what cannot be patched.** Plan and budget to replace end-of-life equipment.
    Where that is not possible yet, isolate it.
 3. **Fix credentials.** Remove default passwords, enforce strong unique credentials
    and add MFA to remote access portals.
 4. **Segment the network.** Keep OT and IoT devices away from corporate IT and the
    internet. Our guide to **_[securing simple office networks](https://secithub.com/unmanaged-switch-security/)_**
    covers the basics.
 5. **Patch faster.** Shrink the time between a vendor fix and deployment, because 
    AI shrinks the time to exploit.
 6. **Watch the traffic.** Grieco notes that network activity does not lie. Unusual
    commands or connections to a controller are a clear warning sign.

    ```wp-block-verse
    If it is connected and cannot be patched, it is not equipment. It is an open door.
    ```

The hacked fridges were not a sophisticated attack. That is exactly why they matter.
Vintage infrastructure hides in every sector, and AI now helps attackers find it
faster than owners can replace it.

The fix is not glamorous. It is inventory, segmentation, strong credentials and 
a real budget for retiring old systems. Organizations that treat those basics as
urgent will spend far less than those who wait for their own overnight surprise.

Want to see how a single weak link can ripple across critical services?

**_[Read our analysis: US Tech Outages | Why a Week of Failures Was Not a Cyber Attack.](https://secithub.com/us-tech-outages-not-cyber-attack/)_**

    ```wp-block-verse
    Retire your vintage tech before an attacker puts it to work.
    ```

![SECITHUB FAQ banner for vintage infrastructure risks, hacked supermarket fridges
and AI-driven attacks](https://secithub.com/wp-content/uploads/2025/12/SECITHUB-
FAQ-1.jpg)

**What is vintage infrastructure?**

Hardware or software that has reached end of life and no longer receives security
updates from the vendor, but is still in use.

**How were European supermarket fridges hacked?**

Attackers reached remote refrigeration controllers and changed temperature settings.
Cisco’s Anthony Grieco says they used known weaknesses, not new techniques.

**Which countries were affected by the fridge attacks?**

Stores in Belgium, Estonia, Ireland, Finland, Luxembourg, the Netherlands and Sweden
reported problems in early November 2026, after an October attack in France.

**Who was behind the supermarket fridge attacks?**

No group has been named, and investigators have not confirmed a link between the
October and November incidents.

**How does AI make old systems more dangerous?**

AI helps attackers scan for exposed devices, match them to known flaws and automate
attacks far faster than before.

**How can organizations protect vintage infrastructure?**

Inventory connected devices, retire or isolate unsupported systems, remove default
passwords, segment networks, patch faster and monitor traffic.

### **Sources**

_[Hacked fridges show the cost of vintage infrastructure – Euronews](https://www.euronews.com/2026/11/18/hacked-fridges-show-the-cost-of-vintage-infrastructure-and-ai-makes-it-worse-says-cisco-ex)_

_[Hackers disrupt refrigeration systems at European supermarkets – BleepingComputer](https://www.bleepingcomputer.com/news/security/hackers-disrupt-refrigeration-systems-at-european-supermarkets/)_

The post [Vintage Infrastructure | Why Hacked Supermarket Fridges Are a Warning for Critical Systems in the AI Era](https://secithub.com/vintage-infrastructure-hacked-fridges/)
appeared first on [SECITHUB](https://secithub.com).

## GPT-6 Cyber Finds 35 Apple Flaws | How AI Broke Through Lockdown Mode

- Link: https://secithub.com/gpt-6-cyber-apple-vulnerabilities/
- Published: 2026-11-18T15:21:19+02:00
- Author: SECITHUB

Reading Time:   5 min

**GPT-6 Cyber** has turned its attention to Apple. According to a Forbes report 
on November 17, 2026, OpenAI and security firm Calif used the new AI model to find
about 35 vulnerabilities across iPhone, Mac, iPad, Apple Watch, Apple TV and Vision
Pro. Some were serious enough to let an attacker take over a device through a malicious
website. Researchers also used the model to write an exploit that strips away protections
in Lockdown Mode, Apple’s highest-security setting.

Apple has already patched many of the flaws, and there is no sign that criminals
exploited them. Still, the findings mark a turning point. AI models now find serious
bugs in one of the most hardened consumer platforms on the market, and they do it
at a pace human teams cannot match.

## **What GPT-6 Cyber Found in Apple’s Platforms**

The research ran over several months. It covered nearly every Apple operating system,
not just the iPhone.

### **About 35 flaws of varying severity**

The bugs ranged from minor issues to remote code execution. The most dangerous ones
could start from a web page. A user would only need to open a malicious link for
an attacker to gain a foothold.

### **Apple’s own credits tell part of the story**

Apple’s security notes now regularly credit AI labs. One tally of Apple advisories
since mid-September counts five CVEs credited to OpenAI and 17 to Calif. Some Calif
entries also credit Anthropic’s Claude. The iOS 27.2 update in October alone fixed
more than 160 vulnerabilities, with credits to OpenAI, Anthropic and Calif.

### **No known attacks in the wild**

Finding a flaw is not the same as someone using it. Apple has not reported active
exploitation of these bugs. That gives users a window to update before criminals
catch up.

    ```wp-block-verse
    Lockdown Mode was built for the most targeted people on Earth. An AI just helped break it.
    ```

## **Why the Lockdown Mode Bypass Matters**

Lockdown Mode is not a normal setting. Apple designed it for journalists, activists,
diplomats and executives who face mercenary spyware such as Pegasus.

### **Built to shrink the attack surface**

When users switch it on, Lockdown Mode blocks most message attachments, limits web
technologies and restricts incoming connections. The goal is simple: fewer features,
fewer ways in. Apple said as recently as spring 2026 that no device with Lockdown
Mode enabled had ever been hacked.

### **Calif had already shown a crack**

In early November, Calif posted a demo of an iPhone 18 Pro compromised with Lockdown
Mode switched on. The attack used a malicious PDF sent over AirDrop and needed the
user to interact with it. The compromise even appeared to survive an update to iOS
27.2. Calif withheld technical details and said it would not release a public jailbreak
tool.

### **Hardening is not immunity**

Lockdown Mode still raises the cost of an attack a great deal. The bypass does not
make it useless. It does show that even the strictest mode depends on code that 
can contain bugs, and AI now hunts for those bugs at scale.

### **Who should keep using it**

Lockdown Mode remains the right choice for people who face targeted threats. That
includes reporters on sensitive beats, human rights workers, officials, executives
who handle deals and security staff with high privileges. Most everyday users do
not need it, because it breaks some features and websites. For them, fast updates
matter far more than any single setting.

## **GPT-6 Cyber and the New Speed of Bug Hunting**

Apple is not an isolated case. The same research team keeps producing major findings,
one after another.

### **A specialist model for security work**

OpenAI launched GPT-6 Cyber in early November as a model built to find, verify and
fix software vulnerabilities. Because of the risk of misuse, the company limits 
access to vetted security partners rather than the public.

### **Calif’s run of findings**

Calif used AI to uncover decade-old PHP flaws in October. This week it disclosed**_
[Java4Shell, a critical Java runtime bug](https://secithub.com/java4shell-cve-2026-114938/)_**
found in about an hour. Apple is the third major platform in roughly a month. The
pattern is clear: AI-assisted research now scales faster than many vendors can patch.

### **The double edge**

Today, responsible researchers and vendors hold the best AI tools. That will not
last. Similar models will reach attackers, and the time between a patch and a working
exploit will keep shrinking. Defenders should plan for that now.

## **What Apple Users and IT Teams Should Do Now**

The fixes exist. The risk lies in devices that have not received them yet. In most
companies, that gap is wider than IT teams think. Personal phones used for work,
old Macs in meeting rooms and Apple TVs in lobbies often fall weeks behind. Each
one is a device an attacker can reach with a single link.

 1. **Update every device.** Install the latest version of iOS, iPadOS, macOS, watchOS,
    tvOS and visionOS. Version 27.2.1 is current as of mid-November.
 2. **Turn on automatic updates.** Let security fixes install without waiting for a
    manual check.
 3. **Enforce versions in MDM.** IT teams should set a minimum OS version and block
    devices that fall behind.
 4. **Keep Lockdown Mode on for high-risk users.** It still blocks far more attacks
    than it lets through.
 5. **Treat links and AirDrop with care.** Set AirDrop to “Contacts Only” and avoid
    opening files from unknown senders.
 6. **Speed up your patch cycle.** AI-found bugs mean more fixes, more often. Our guide
    to **_[defending against today’s top cyberattacks](https://secithub.com/tactics-tools-cyberattacks-2025/)_**
    covers fast patching habits.

    ```wp-block-verse
    The best defense against AI-found bugs is still the oldest one: update fast.
    ```

GPT-6 Cyber did not hack anyone. It helped responsible researchers find flaws before
criminals did, and Apple fixed many of them. That is the system working as it should.

But the speed is new. When AI tools can surface dozens of bugs in Apple, Java and
PHP within weeks, every vendor’s patch pipeline is under pressure. Users and IT 
teams who update quickly will stay ahead. Those who wait will soon face attackers
with the same tools.

Want to understand how AI agents are changing both attack and defense?

**_[Read our deep-dive: AI Agents Between Hype and Reality.](https://secithub.com/ai-agents-2025-hype-vs-reality/)_**

    ```wp-block-verse
    Patch before the attackers get the same AI you just read about.
    ```

![SECITHUB FAQ banner for GPT-6 Cyber Apple vulnerabilities and the Lockdown Mode
bypass](https://secithub.com/wp-content/uploads/2025/12/SECITHUB-FAQ-1.jpg)

**What did GPT-6 Cyber find in Apple products?**

About 35 vulnerabilities across iPhone, Mac, iPad, Apple Watch, Apple TV and Vision
Pro, according to Forbes. Some could allow device takeover through a malicious website.

**Was Apple’s Lockdown Mode really bypassed?**

Researchers used GPT-6 Cyber to write an exploit that removes some Lockdown Mode
protections. Calif had earlier shown a separate bypass that needed user interaction.

**Have these Apple vulnerabilities been exploited?**

There are no reports of active exploitation. Apple has already fixed many of the
flaws.

**Who found the vulnerabilities?**

OpenAI and the security firm Calif, using the GPT-6 Cyber model over several months.

**Is GPT-6 Cyber available to the public?**

No. OpenAI limits access to vetted security partners because of the risk of misuse.

**What should Apple users do?**

Install the latest updates on every Apple device, turn on automatic updates and 
keep Lockdown Mode on if you face targeted threats.

### **Sources**

_[Calif.io iOS attack defeats Lockdown Mode and survives update – heise online](https://www.heise.de/en/news/Calif-io-iOS-attack-defeats-Lockdown-Mode-and-survives-update-11511465.html)_

_[iOS 27.2 Fixes Over 160 Security Vulnerabilities – MacRumors](https://macrumors.com/2026/10/26/ios-27-2-security-fixes)_

_[Apple security releases – Apple Support](https://support.apple.com/en-us/100100)_

The post [GPT-6 Cyber Finds 35 Apple Flaws | How AI Broke Through Lockdown Mode](https://secithub.com/gpt-6-cyber-apple-vulnerabilities/)
appeared first on [SECITHUB](https://secithub.com).

## US Tech Outages | Why a Week of Failures Was Not a Cyber Attack, and Why That Should Still Worry You

- Link: https://secithub.com/us-tech-outages-not-cyber-attack/
- Published: 2026-11-18T15:16:52+02:00
- Author: SECITHUB

Reading Time:   5 min

A week of **US tech outages** has left many Americans asking the same question: 
is someone attacking the country’s digital backbone? Microsoft 365 went dark, the
FAA grounded every departing flight, Chase customers struggled to move money and
Apple services stumbled. On November 17, Secretary of State Marco Rubio said the
US is “not under systemic attack” and that the events are unrelated.

Every company and agency involved has said the same: no hacker, no hostile state,
just technical failures. For security teams, that answer is not the end of the story.
It may be the more worrying one. A string of ordinary mistakes did damage that many
cyber attacks never achieve.

## **A Week of US Tech Outages, Day by Day**

Here is what failed, and what each organization says caused it.

### **Thursday: Microsoft 365 goes down**

Outlook, Teams, SharePoint and Copilot failed for users worldwide for nearly 24 
hours, from November 12 to November 13. Microsoft blamed three factors: a hidden
bug in its networking gateway servers, a sudden surge in traffic and reduced server
capacity. The company said malicious activity played no part.

### **Friday: the FAA stops every flight**

On November 13, a communications failure hit air traffic control centers across 
the country. The FAA halted departures nationwide for more than four hours. It was
only the second nationwide ground stop since 9/11. Officials pointed to overloaded
network equipment and ageing infrastructure. Homeland Security said it found no 
sign of a cyber attack.

### **Monday: banks and airlines**

A technical problem at JPMorgan Chase slowed transactions for its own customers 
and for customers at other banks that deal with it. Chase said the cause was not
a cyber attack. United Airlines also reported a software failure that delayed flights
the same day.

### **Tuesday: Apple and a classified network**

Apple users lost access to iCloud and Apple Pay for a time. Apple said an internal
update triggered the problem. Reports also said an internal NSA communications network
went offline briefly.

    ```wp-block-verse
    Five failures in five days, and not one attacker. That should worry us more, not less.
    ```

## **Why Officials Say It Is Not an Attack**

Rubio said every indicator points to separate causes with no link to criminals or
foreign states. He added that sharing what the government knows is the best answer
to public fear. The Director of National Intelligence made a similar point, saying
many institutions are patching and upgrading systems at once.

### **Different systems, different causes**

The failures hit unrelated platforms run by different operators. Microsoft’s problem
sat in cloud gateway servers. The FAA’s sat in telecom and routing gear. Chase and
United cited software bugs. A coordinated attack usually leaves shared fingerprints,
such as the same malware, the same entry point or the same timing. None has surfaced
so far.

### **Experts agree, with a caveat**

Independent security researchers have also found no evidence that ties the incidents
together. Still, “no evidence” is a statement about what investigators know today.
Full post-incident reports from the FAA, Apple and the banks have not yet appeared.
Healthy skepticism means waiting for those, not filling the gap with theories.

### **Fear spreads faster than facts**

Social media filled quickly with claims of a coordinated plot. Reporters even asked
Rubio about the risk of bank runs or panic buying. Attackers also exploit moments
like this. Phishing emails that pose as outage notices or “account verification”
requests tend to spike right after big disruptions.

## **What the US Tech Outages Really Reveal**

If the official account holds, the lesson is not about hackers. It is about how 
fragile tightly linked systems have become.

### **Change is the biggest risk**

Most of these failures followed a change: a network upgrade, a software update or
new equipment. Much core infrastructure is decades old, and teams now have to modernize
it fast while it keeps running. Every change carries a chance of breaking something
nobody fully understands.

### **Concentration multiplies impact**

One bug at Microsoft cut off millions of workers at once. One failure in FAA telecom
stopped flights in every state. When so much depends on so few platforms, a single
fault becomes a national event. We saw the same pattern in **_[Cloudflare’s November 2025 outage](https://secithub.com/cloudflare-outage-2025-analysis/)_**,
when one bad configuration file broke a large part of the web.

### **Availability is a security issue**

Security teams often focus on keeping attackers out. But the classic security triad
includes availability, not just confidentiality and integrity. A system that is 
down hurts the business whether a hacker or a bug took it offline.

## **What Organizations Should Do Now**

You cannot prevent a Microsoft or FAA outage. You can decide how much it hurts you.

 1. **Map your critical dependencies.** List the cloud, identity, payment and communication
    services your business cannot run without.
 2. **Plan for each one going dark.** Write down how your team works if email, chat
    or payments fail for a full day.
 3. **Add backup channels.** Keep a second way to reach staff and customers that does
    not depend on your main provider.
 4. **Treat your own changes as risk.** Use staged rollouts, clear rollback plans and
    change freezes before busy periods.
 5. **Warn staff about outage phishing.** Remind people that real providers rarely 
    ask for passwords during an incident.
 6. **Rehearse the “is it an attack?” call.** Decide in advance who confirms the cause
    and what you tell customers in the first hour. Our guide to **_[building cyber resilience](https://secithub.com/cyber-resilience-hosting-smbs-2025/)_**
    covers the basics.

    ```wp-block-verse
    Resilience means your business keeps working when someone else's system does not.
    ```

The week of US tech outages will likely end up as a list of separate post-mortems,
not a single plot. That is good news for national security. It is also a warning.
Modern infrastructure can fail at national scale without any help from an adversary.

Attackers study weeks like this one too. They now know how much disruption a single
weak point can cause. The organizations that plan for outages today will be ready
whether the next one comes from a bug or a breach.

Want to see how third-party and cloud dependencies shape your real attack surface?

**_[Read our deep-dive: The New Perimeter is the Supply Chain | Managing Third-Party and SaaS Risk.](https://secithub.com/modern-cyber-defense-strategies-2026/)_**

    ```wp-block-verse
    Plan for the outage you cannot prevent before it plans your week for you.
    ```

![SECITHUB FAQ banner for the week of US tech outages covering Microsoft, FAA, Chase
and Apple failures and infrastructure resilience](https://secithub.com/wp-content/
uploads/2025/12/SECITHUB-FAQ-1.jpg)

**Were the US tech outages caused by a cyber attack?**

Officials and every affected company say no. Each outage has been linked to technical
causes such as bugs, updates or equipment failures.

**Which services failed during the week of US tech outages?**

Microsoft 365, FAA air traffic communications, JPMorgan Chase transactions, United
Airlines systems and Apple iCloud and Apple Pay.

**Are the outages connected to each other?**

No link has been found. They hit different systems run by different operators, with
different reported causes.

**What caused the FAA ground stop?**

The FAA cited a communications failure across air traffic control centers, tied 
to overloaded network equipment. Homeland Security said it found no sign of an attack.

**Why do so many outages happen at once?**

Many organizations are upgrading old infrastructure quickly, and every change carries
risk. Heavy reliance on a few big platforms also magnifies each failure.

**How can businesses prepare for similar outages?**

Map critical dependencies, plan for each one failing, add backup communication channels,
manage changes carefully and warn staff about outage-themed phishing.

### **Sources**

_[Nationwide Ground Stop: Questions and Answers – FAA](https://www.faa.gov/newsroom/nationwide-ground-stop-questions-and-answers)_

_[Microsoft Explains Last Week’s 365 Outage – TechPowerUp](https://www.techpowerup.com/354904/microsoft-explains-last-weeks-365-outage)_

_[A technical issue at JPMorgan Chase triggers online banking delays – ABC News](https://abc11.com/post/customers-nations-largest-banks-are-reporting-difficulty-accessing-accounts-online/20045188/)_

The post [US Tech Outages | Why a Week of Failures Was Not a Cyber Attack, and Why That Should Still Worry You](https://secithub.com/us-tech-outages-not-cyber-attack/)
appeared first on [SECITHUB](https://secithub.com).

## Microsoft Agent Framework PyPI Compromise | How a 55-Minute Release Stole Cloud Secrets

- Link: https://secithub.com/microsoft-agent-framework-pypi-compromise/
- Published: 2026-11-18T15:09:35+02:00
- Author: SECITHUB

Reading Time:   5 min

A malicious **Microsoft Agent Framework PyPI** release went live on the evening 
of November 17, 2026. Two poisoned versions of the `agent-framework` package, 1.26.1
and 1.26.2, sat on the Python Package Index for about 55 minutes. In that window,
any developer laptop, CI runner or container that pulled them ran a credential stealer
aimed squarely at Azure, GitHub and AI developer secrets.

Wiz researchers caught and analyzed the release. The package sees about 41,000 downloads
a week, so even a short window matters. It also follows several PyPI attacks this
year that used the same trick: a startup hook that runs before your code does.

## **What Happened to the Microsoft Agent Framework PyPI Package**

Microsoft Agent Framework is the company’s open-source toolkit for building AI agents
in Python and .NET. Its PyPI package is a common dependency in agent projects and
Azure AI pipelines.

### **A 55-minute window**

Version 1.26.1 appeared at 20:35 UTC. A second malicious build, 1.26.2, followed
around 21:08 UTC. PyPI quarantined both by about 21:30 UTC. The last clean release
is 1.26.0, published on November 12.

### **Built outside the normal pipeline**

The Microsoft GitHub repository has no tag or release for 1.26.1, and its main branch
holds no malicious code. Legitimate releases ship a wheel, a source distribution
and a matching `agent-framework-core` version. The bad release shipped a lone wheel.
Everything points to a direct upload with stolen publishing access, although the
root cause is not yet confirmed.

### **A second version that hid better**

The attacker did not stop at one build. Version 1.26.2 renamed its files, restored
a clean `__init__.py` and widened its target list. Teams that searched only for 
the 1.26.1 file names would miss it.

    ```wp-block-verse
    Fifty-five minutes is all a stolen publishing token needs.
    ```

## **How the Stealer Works**

The payload is simple, quiet and fast. It aims to grab as many secrets as possible
in one shot.

### **A .pth file runs it on every Python start**

Each version drops a `.pth` file into `site-packages`. Python executes these files
every time the interpreter starts. So the stealer runs even if your code never imports`
agent-framework`. Any Python process in an affected environment triggered it.

### **It harvests secrets from everywhere**

 * Environment variables with names like TOKEN, SECRET, KEY, AZURE, AWS, GITHUB,
   OPENAI and HF_
 * Secret values from up to 300 other running processes
 * Up to 80 credential files, including Azure CLI config, `.npmrc`, `.git-credentials`,
   Docker config and Hugging Face tokens
 * Azure managed identity tokens from the instance metadata service
 * Azure DevOps and GitHub Actions OIDC tokens that can be swapped for Azure access

### **One quiet upload**

The stealer packs everything into one compressed JSON blob and sends it in a single
HTTPS POST to a webhook.site URL. File and function names mimic SDK telemetry, so
a quick glance looks harmless. It does not retry on failure and leaves little noise
behind.

### **A small catch, a big risk**

The attacker used a free webhook.site account, which caps captures at 100 requests.
Wiz says no more than 50 requests landed. That sounds small, but a single CI runner
can hold keys to an entire cloud tenant.

## **Why AI Developer Tooling Is Now a Prime Target**

This attack fits a clear 2026 pattern. Attackers go after the packages that AI developers
trust most, because those environments hold the richest secrets.

### **Microsoft packages have been hit before**

In May 2026, attackers used a stolen PyPI token to push three trojanized versions
of Microsoft’s `durabletask` SDK. Wiz has not linked the two incidents. Still, a
second Microsoft Python package hit through the same kind of direct upload raises
hard questions about publishing controls.

### **Startup hooks are the new favorite**

In June, a wave of poisoned wheels known as Hades spread through 19 PyPI packages
using `.pth` hooks. In October, a compromised LiteLLM release used the same Python
startup trick to steal credentials. Defenders often scan imports and entry points.
A `.pth` file slips past that, because it runs before any of your code loads.

### **Agent builders hold the keys**

Teams that build AI agents keep model API keys, cloud identities and deployment 
tokens close at hand. Their pipelines often have broad Azure rights. That makes 
them a far better haul than an average web app. Our guide to **_[securing AI bots](https://secithub.com/secure-ai-bot-2025/)_**
covers why these secrets need tighter scope.

### **Provenance checks help, but only if you use them**

The bad release broke every pattern of a normal Microsoft build. It had no Git tag,
no source distribution and no matching core package. Tools that compare a release
against its repository and build history would have flagged it at once. Few teams
run those checks before they install.

## **What to Do Now**

If any environment you run installed `agent-framework` between 20:35 and 21:30 UTC
on November 17, treat it as compromised.

 1. **Find exposure.** Search lockfiles, CI logs, virtual environments and container
    images for versions 1.26.1 or 1.26.2. Pin to 1.26.0.
 2. **Check for the files.** Look in `site-packages` for `agent_framework_meta.pth`
    or `af_runtime_diag.pth`, and for the `_AF_TEL=1` variable.
 3. **Check egress.** Search proxy and DNS logs for `webhook.site/d381fd80-ec62-42ad-
    8f5c-d1b15103c9bb`.
 4. **Rotate everything.** Replace environment secrets, Azure CLI sessions, Azure DevOps
    tokens, git credentials, and npm, PyPI, Docker and Hugging Face tokens on affected
    hosts.
 5. **Audit Azure identities.** Review sign-in logs for managed identities and federated
    credentials. Stolen tokens may stay valid for up to 24 hours.
 6. **Harden your pipeline.** Pin dependencies with hashes, delay new versions by a
    few days, and give CI jobs only the cloud rights they need. Our **_[supply chain security guide](https://secithub.com/supply-chain-security-guide-2025/)_**
    walks through these controls.

    ```wp-block-verse
    Pin your versions, scope your tokens and assume the next poisoned release is already on its way.
    ```

The Microsoft Agent Framework PyPI compromise lasted less than an hour, yet it could
reach deep into cloud environments. A trusted vendor name and a familiar package
were enough to put a stealer in front of thousands of builds.

The fix is not to stop trusting open source. It is to stop trusting it blindly. 
Hash pinning, release delays and least-privilege CI tokens turn a 55-minute attack
into a non-event.

Want to see how attackers move through partners, packages and SaaS links long before
anyone notices?

**_[Read our deep-dive: The New Perimeter is the Supply Chain | Managing Third-Party and SaaS Risk.](https://secithub.com/modern-cyber-defense-strategies-2026/)_**

    ```wp-block-verse
    Lock down your build pipeline before the next poisoned package lands in it.
    ```

![SECITHUB FAQ banner for the Microsoft Agent Framework PyPI compromise covering
malicious versions, credential theft and remediation](https://secithub.com/wp-content/
uploads/2025/12/SECITHUB-FAQ-1.jpg)

**Which Microsoft Agent Framework PyPI versions are malicious?**

Versions 1.26.1 and 1.26.2 of the agent-framework package. The last safe version
is 1.26.0.

**How long were the malicious versions available?**

About 55 minutes on November 17, 2026, from 20:35 UTC until PyPI removed them around
21:30 UTC.

**Do I need to import the package to be affected?**

No. A .pth file runs on every Python start, so any Python process in an affected
environment ran the stealer.

**What did the malware steal?**

Secrets from environment variables, credential files, other processes, Azure managed
identities and CI OIDC tokens.

**How do I check if I was hit?**

Search for versions 1.26.1 or 1.26.2, look for agent_framework_meta.pth or af_runtime_diag.
pth, and check logs for the webhook.site URL.

**What should I do if I installed a bad version?**

Pin to 1.26.0, rotate every secret on the affected host or runner, and audit Azure
sign-in logs for the exposed identities.

### **Sources**

_[Malicious version of Microsoft’s Agent Framework published to PyPI – Wiz](https://www.wiz.io/blog/malicious-version-of-microsofts-agent-framework-published-to-pypi)_

_[Inside the LiteLLM 1.107.0 supply chain compromise – Microsoft Security](https://microsoft.com/en-us/security/blog/2026/10/23/inside-the-litellm-1-107-0-supply-chain-compromise-credential-theft-through-python-startup-hooks)_

The post [Microsoft Agent Framework PyPI Compromise | How a 55-Minute Release Stole Cloud Secrets](https://secithub.com/microsoft-agent-framework-pypi-compromise/)
appeared first on [SECITHUB](https://secithub.com).

## Hertz CCPA Fine | How Ignored Opt-Out Signals Cost $1.55 Million

- Link: https://secithub.com/hertz-ccpa-fine/
- Published: 2026-11-18T15:04:32+02:00
- Author: SECITHUB

Reading Time:   5 min

The **Hertz CCPA fine** landed on November 17, 2026, when the California Privacy
Protection Agency (CalPrivacy) ordered the rental car giant to pay $1.55 million
and overhaul its privacy practices. Regulators say Hertz ignored browser opt-out
signals, hid how it handled them and made opting out harder than it should be. Meanwhile,
trackers on its websites and apps kept sending rental search data to ad tech platforms.

The case is not a data breach. No hacker stole anything. Yet it may matter more 
to most companies than many breach headlines, because the failures behind it are
common. If your site runs Meta, Google or TikTok pixels, this order describes risks
you probably share.

## **What Led to the Hertz CCPA Fine**

CalPrivacy’s Enforcement Division brought four main claims. The conduct ran from
July 2023 to August 2026 and covered the Hertz, Dollar and Thrifty brands.

### **Opt-out signals went unprocessed**

Under the CCPA, a business must treat a browser opt-out preference signal, such 
as Global Privacy Control (GPC), as a valid request to stop selling or sharing data.
According to the regulator, Hertz did not honor those signals. Visitors who had 
switched GPC on still had their data shared for advertising.

### **Trackers fed ad platforms**

Pixels and SDKs on the three brands’ websites and apps sent rental searches and 
device identifiers to ad tech firms, including Meta, Google and TikTok. That kind
of transfer counts as “sharing” for cross-context advertising under California law.

### **A cookie banner that undid itself**

On Dollar.com and Thrifty.com, users could switch off advertising cookies in a “
Manage Cookies” window. But if they closed that window with the X icon instead of
clicking “Save Changes,” the cookies switched back on. Regulators called this unnecessary
friction.

### **No clear disclosure**

The company also failed to explain how it handled opt-out preference signals. The
CCPA requires businesses to state this in their privacy policy.

    ```wp-block-verse
    If closing a cookie banner turns tracking back on, the banner is not consent. It is decoration.
    ```

## **What Hertz Must Change**

The money is only part of the order. The operational demands will cost more effort
than the fine itself.

 * Build and maintain a full inventory of trackers on its sites and apps
 * Review every new tracker before it goes live
 * Test opt-out handling every quarter across staging and production, browsers, 
   mobile platforms and each tag management container
 * Report publicly how many opt-outs arrive through preference signals like GPC
 * Keep these obligations in place for three years

Quarterly testing across every tag container is the detail to watch. It shows the
regulator expects proof that opt-outs work in practice, not just a policy that says
they should.

### **Why public reporting matters**

The duty to publish opt-out numbers is unusual. It turns a private fix into a public
scorecard. Customers, journalists and rivals will see how many people use GPC on
Hertz sites, and whether those numbers look plausible. A sudden drop or a suspiciously
low count would invite fresh questions. For other companies, it hints at where future
orders may go: less trust in promises, more demand for measurable proof.

## **Why the Hertz CCPA Fine Fits a Bigger Pattern**

Hertz is not an outlier. California has turned opt-out failures into its main enforcement
theme.

### **A busy month for CalPrivacy**

One week earlier, the agency announced a fine of close to $3 million against FedEx
over CCPA violations. On the same day as Hertz, it ordered outdoor retailer C.C.
Filson to stop all device-based selling and sharing of data with ad tech firms within
30 days. Filson must stay suspended until it proves compliance, and it will pay 
$217,987 after showing it could not afford a larger fine.

### **Fines keep climbing**

In 2025, Tractor Supply paid $1.35 million, then a record for the agency. The state
attorney general settled with Healthline for $1.55 million the same year. In 2026,
Ford, PlayOn Sports and others faced orders focused on opt-out rights. Each case
raised the bar on what regulators expect.

### **“Compliance theater” no longer works**

A cookie banner and a privacy policy are no longer enough. Regulators now test whether
opt-outs actually stop data flows. They load pages with GPC switched on and watch
which trackers still fire. Any company can run the same test today.

### **Privacy and security now overlap**

Third-party scripts are both a privacy risk and a security risk. Each pixel is outside
code running on your site, and it can change without notice. That is why tracker
governance belongs next to your **_[software supply chain security](https://secithub.com/supply-chain-security-guide-2025/)_**
program, not only in the legal team’s inbox.

## **What Companies Should Do Now**

These steps apply to any business with California visitors, whether it rents cars
or sells software.

 1. **Test with GPC on.** Turn on Global Privacy Control in a browser, load your key
    pages and check which trackers still send data.
 2. **Inventory every tracker.** List each pixel, SDK and tag, who owns it and what
    data it sends. Include mobile apps.
 3. **Fix your banner logic.** Make sure closing a window never re-enables cookies.
    Every exit path should respect the user’s last choice.
 4. **Gate new tags.** Require privacy and security review before anyone adds a tracker
    to a tag manager.
 5. **Document signal handling.** State in your privacy policy how you process GPC 
    and similar signals.
 6. **Test on a schedule.** Adopt quarterly checks like those in the Hertz order before
    a regulator imposes them. Our guide on **_[modern GRC](https://secithub.com/grc-ai-resilience-2025/)_**
    shows how to build this into governance.

    ```wp-block-verse
    Regulators now test your opt-outs the way attackers test your login page.
    ```

The Hertz CCPA fine is not about a dramatic breach. It is about small technical 
gaps that quietly broke a legal promise for three years. A missed signal, a banner
that reset itself and pixels nobody reviewed added up to a seven-figure penalty.

The good news is that these gaps are easy to find. Any team can test them in an 
afternoon. The cost of not looking keeps rising with every new order from Sacramento.

Want to turn compliance from a checkbox into real business value?

**_[Read our deep-dive: ISO 27001 in 2025 | From Certification to Real ROI.](https://secithub.com/iso-27001-2025-compliance-roi/)_**

    ```wp-block-verse
    Audit your trackers before California audits them for you.
    ```

![SECITHUB FAQ banner for the Hertz CCPA fine covering Global Privacy Control, opt-
out failures and tracker compliance](https://secithub.com/wp-content/uploads/2025/
12/SECITHUB-FAQ-1.jpg)

**How much is the Hertz CCPA fine?**

CalPrivacy ordered Hertz to pay $1.55 million and change its privacy practices for
three years.

**Why was Hertz fined?**

Regulators say Hertz ignored Global Privacy Control opt-out signals, did not disclose
how it handled them and made opting out harder than necessary.

**Which Hertz brands were involved?**

The case covered the Hertz, Dollar and Thrifty websites and apps. The cookie banner
problem appeared on Dollar and Thrifty.

**What is Global Privacy Control?**

GPC is a browser setting that tells websites not to sell or share your data. California
law requires businesses to honor it.

**Was this a data breach?**

No. The case concerns how Hertz shared data with ad tech firms through trackers,
not a hack or theft of data.

**How can companies avoid a similar fine?**

Test pages with GPC on, inventory every tracker, fix banner logic, review new tags
and test opt-outs on a regular schedule.

### **Sources**

_[Hertz to Pay $1.55 Million CCPA Fine, Adjust Privacy Practices – California Privacy Protection Agency](https://privacy.ca.gov/2026/11/vehicle-rental-services-company-hertz-to-pay-1-55-million-ccpa-fine-adjust-privacy-practices/)_

_[Hertz decision – California Privacy Protection Agency](https://privacy.ca.gov/2026/11/hertz/)_

The post [Hertz CCPA Fine | How Ignored Opt-Out Signals Cost $1.55 Million](https://secithub.com/hertz-ccpa-fine/)
appeared first on [SECITHUB](https://secithub.com).

## The Gentlemen Ransomware Claims NVIDIA | Why a Leak-Site Listing Is Not Proof of a Breach

- Link: https://secithub.com/the-gentlemen-ransomware-nvidia/
- Published: 2026-11-18T15:01:35+02:00
- Author: SECITHUB

Reading Time:   5 min

**The Gentlemen ransomware** group added NVIDIA to its dark web leak site on November
18, 2026. So far, the listing is only a claim. The gang has posted no data sample,
no file list and no proof of access, and NVIDIA has not confirmed any breach. Still,
a claim against the world’s most valuable chipmaker deserves a close look, because
the group behind it is now one of the busiest ransomware operations on the planet.

This article covers what the listing shows, who The Gentlemen are, how to judge 
a leak-site claim, and what security teams should take from it. We will update it
if NVIDIA responds or the group publishes evidence.

## **What The Gentlemen Ransomware Group Claims**

Threat intelligence trackers spotted the new entry on the morning of November 18.
The listing is thin, which matters for how seriously to take it.

### **A name, a ticker and a timer**

The post names NVIDIA, its revenue and its NVDA stock ticker. It also shows a countdown
of roughly ten days before a threatened leak. A Tox chat ID invites the company 
to “negotiate.”

### **No evidence yet**

The data section simply says “SOON.” The group has not shared screenshots, file 
trees or sample documents. Researchers who track leak sites mark the claim as pending
verification.

### **Silence from NVIDIA**

NVIDIA has not commented publicly. That silence proves nothing either way. Large
companies often investigate quietly before they say anything about an extortion 
claim.

    ```wp-block-verse
    A leak-site listing is a sales pitch, not a forensic report.
    ```

## **Who Is Behind The Gentlemen Ransomware?**

The Gentlemen are not a fringe crew chasing headlines. In under eighteen months,
the group has grown into a top-tier ransomware-as-a-service (RaaS) brand.

### **Born from a Qilin split**

Before launching its own brand, the core team worked as ArmCorp, a prolific affiliate
of the Qilin ransomware program. In July 2025, its leader, known as “hastalamuerte,”
publicly accused Qilin of withholding about $48,000 in commissions. Researchers 
at Group-IB found the first Gentlemen sample days before that dispute went public.
The split looks planned, not impulsive.

### **A Russian-speaking RaaS business**

The operators rent their ransomware to affiliates, who carry out the intrusions.
Affiliates reportedly keep up to 90% of each ransom, an aggressive offer that pulls
talent from rival programs. The group bans attacks on Russia and other CIS countries,
a common sign of a Russian-speaking operation. Microsoft tracks the operators as
Storm-2697.

### **Double extortion at scale**

Affiliates steal data first, then encrypt systems, then threaten to publish the 
stolen files if the victim does not pay. According to Check Point Research, The 
Gentlemen were the most active ransomware group in October 2026, behind 13% of all
publicly claimed attacks. Qilin came second with 12%.

### **An unusual target for this group**

The Gentlemen rarely go after US giants. In Check Point’s first-quarter 2026 data,
only 13% of the group’s victims sat in the United States, against an ecosystem average
near 50%. Most of its activity clustered in Asia-Pacific and Latin America. Industrial
firms are a favorite too. Dragos counted 167 industrial victims from the group in
the third quarter alone. A listing for a US tech leader like NVIDIA breaks that 
pattern, which is one more reason to wait for proof.

### **Fast, cross-platform tooling**

Microsoft describes the encryptor as Go-based malware that pairs strong per-file
encryption with aggressive self-propagation. It runs on Windows, Linux, NAS devices
and VMware ESXi. Documented intrusions include abuse of the FortiGate authentication
bypass CVE-2024-55591 and ransomware pushed through malicious Group Policy Objects.

## **How to Read a Ransomware Leak-Site Claim**

Ransomware groups use leak sites as marketing. A famous name boosts their brand,
attracts affiliates and pressures other victims to pay. That gives gangs a reason
to exaggerate.

### **Claims often outrun reality**

Some listings turn out to be real breaches. Others point to a subsidiary, a supplier
or a contractor that holds some company data. A few recycle files from old leaks.
The headline name is not always the network that fell.

### **Signals that raise confidence**

 * Sample files with internal metadata, file paths or recent dates
 * A file tree that matches known company systems
 * Confirmation from the company, a regulator filing or an SEC 8-K
 * Independent researchers who validate the data

Right now, the NVIDIA listing has none of these. Treat it as unverified until that
changes.

### **NVIDIA has been a target before**

In 2022, the LAPSUS$ gang breached NVIDIA and leaked employee credentials and code-
signing certificates. Attackers later used those certificates to sign malware. A
company with that history and that much valuable IP will stay in attackers’ sights,
whether this claim holds up or not.

## **What Security Teams Should Do Now**

You do not need to wait for NVIDIA’s answer to act. The Gentlemen hit hundreds of
organizations every quarter, and their playbook is well documented.

 1. **Patch your edge devices.** Affiliates break in through exposed firewalls and 
    VPNs. Close known flaws such as CVE-2024-55591 on FortiGate first.
 2. **Guard Group Policy.** Alert on new or changed GPOs, and limit who can edit them.
    Our guide to **_[Active Directory security](https://secithub.com/active-directory-security-hardening-identity/)_**
    covers the key controls.
 3. **Protect ESXi hosts.** Isolate management interfaces, enforce MFA and keep offline
    backups of virtual machines.
 4. **Watch for data staging.** Large archives, new cloud sync tools and unusual outbound
    traffic often appear before encryption starts.
 5. **Review supplier access.** Many “big name” breaches start at a partner. Audit 
    vendor accounts and remote access paths.
 6. **Plan your response to a claim.** Decide in advance who verifies a leak-site listing,
    who talks to the press and when legal and regulators get involved.

    ```wp-block-verse
    Verify before you panic, but prepare as if the next listing has your name on it.
    ```

The Gentlemen’s claim against NVIDIA may prove real, partial or empty. The countdown
suggests we will know more within days. Either way, the listing shows how far the
group’s ambitions have grown, from mid-sized manufacturers to the center of the 
AI economy.

For defenders, the lesson is practical. The tools and entry points this group uses
are known. Closing them is far cheaper than responding to a countdown timer.

Want to see how attackers move through partners and SaaS links long before anyone
notices?

**_[Read our deep-dive: The New Perimeter is the Supply Chain | Managing Third-Party and SaaS Risk.](https://secithub.com/modern-cyber-defense-strategies-2026/)_**

    ```wp-block-verse
    Close the doors ransomware affiliates use most before they knock on yours.
    ```

![SECITHUB FAQ banner for The Gentlemen ransomware claim against NVIDIA, covering
the group's background and how to verify leak-site claims](https://secithub.com/
wp-content/uploads/2025/12/SECITHUB-FAQ-1.jpg)

**Did The Gentlemen ransomware group really breach NVIDIA?**

It is not confirmed. The group listed NVIDIA on its leak site but has shown no data,
and NVIDIA has not confirmed a breach.

**Who are The Gentlemen?**

A Russian-speaking ransomware-as-a-service group that launched in mid-2025. Its 
founders previously worked as an affiliate of Qilin.

**How active is The Gentlemen ransomware?**

Very active. Check Point Research ranked it the top ransomware group in October 
2026, with 13% of publicly claimed attacks.

**How do The Gentlemen attack?**

Affiliates break in through exposed edge devices, steal data, then deploy a cross-
platform encryptor that also targets VMware ESXi.

**Why would a gang list a victim without proof?**

Leak sites are marketing. A famous name raises pressure on victims and attracts 
affiliates, so some claims are exaggerated or wrong.

**What should organizations do now?**

Patch edge devices, monitor Group Policy changes, protect ESXi hosts, watch for 
data staging and audit supplier access.

### **Sources**

_[The Gentlemen ransomware: Dissecting a self-propagating Go encryptor – Microsoft Security](https://www.microsoft.com/en-us/security/blog/2026/05/28/the-gentlemen-ransomware-dissecting-a-self-propagating-go-encryptor/)_

_[hastalamuerte and The Gentlemen RaaS – Group-IB](https://www.group-ib.com/blog/hastalamuerte-gentlemen-raas-ttps)_

_[Victim listing: NVIDIA – Ransomware.live](https://www.ransomware.live/id/TlZJRElBQHRoZWdlbnRsZW1lbg)_

The post [The Gentlemen Ransomware Claims NVIDIA | Why a Leak-Site Listing Is Not Proof of a Breach](https://secithub.com/the-gentlemen-ransomware-nvidia/)
appeared first on [SECITHUB](https://secithub.com).
