We use many technologies to detect email spam. They analyze every part of a message: headers, text, images, and links. But the first thing our email products focus on is the “From” field. If the address contains a domain that’s already on our denylist, it’s an easy call. But spammers know the drill, so they register domains in industrial quantities to make emails, even from a single campaign, arrive from completely different addresses. Clearly, they don’t do that by hand. The process has long been automated, and attackers’ algorithms can register huge numbers of domains in a matter of minutes. But every domain needs a name. Coming up with a unique one for each isn’t feasible. So spammers use domain generation algorithms (DGAs), which create similar names with minor variations, based on date and time, pseudo-random numbers, and so on.
In fact, to block mailings such as these, you don’t need every address on your denylist. We’re good at automating processes, too. For some time now, our email solutions have included a technology that detects domains generated by DGAs, and shuts down entire batch with a single rule. Let me explain how this technology works.
How we detect and block generated domain names in bulk
As soon as any of our technologies flags an email as spam, its address becomes subject to further examination. If it meets a number of criteria, its domain is added to our database of domains we consider “bad”. We built a cascade of classifiers that compares these domains by similarity, groups them into clusters, and creates a short but comprehensive rule that blocks the entire cluster. Our email protection solutions receive this rule with an update. From then on, they block all emails sent from addresses on any domain that matches it — even if those domains have never been used in spam campaigns yet.
Here’s a real-world example. Let’s say the classifier flags several lookalike domains in a collection of “bad” ones it has gathered:
sxkjcc.asia
sxkjcd.asia
sxkjcy.asia
Then, one rule is enough to catch them all: @sxkjc[a-z0-9]\.asia\b
How to protect your company from spam
According to our analysis, 44.9% of emails sent last year were spam. Unwanted, unsolicited emails distract employees, waste working time, and clog inboxes. In some cases, they also contain scam, phishing, or malware content. To protect business customers from this deluge of threats, our solutions for corporate email security have anti-spam technologies that we constantly improve. This includes the technology described above for blocking domains generated by DGAs.
Two critical vulnerabilities have been discovered in the popular PHP programming language. The vulnerabilities, identified as CVE-2026-109618 and CVE-2026-109619, allow attackers to remotely execute code on a server running PHP without authentication. The language’s development team has released updates for all supported branches, but the first attempts to exploit the vulnerabilities were observed even before the patches were published. The exploit has been named PHP2Shell.
The architectures of popular marketplaces, e-commerce platforms, and learning management systems (LMS) are written in PHP. In addition, PHP is used in various content management systems (CMS) such as WordPress, Drupal, Joomla, and so on, which power most websites on the internet.
In this blogpost, we discuss CVE-2026-109618 and CVE-2026-109619 in detail, and explain how companies can protect their infrastructure. Read the Securelist expert blog to learn how events surrounding these vulnerabilities unfolded before the patch was released.
Vulnerable PHP versions
According to PHP web site, versions prior to 8.2.35, 8.3.36, 8.4.27, and 8.5.12 are vulnerable to CVE-2026-109618 and CVE-2026-109619. There are no official fixes for older branches. According to unofficial reports, the vulnerabilities also affect PHP versions starting with 5.4.0.
What makes CVE-2026-109618 and CVE-2026-109619 dangerous
Both vulnerabilities received a score of 9.2 on the CVSS v4.0 scale. They arise because, after a request is interrupted (for example, due to a memory shortage), PHP leaves invalid data in the memory buffer, and subsequently accesses it, allowing code to be executed on the server. One of the confirmed attack chains via CVE-2026-109619 exploits the `session.upload_progress` mechanism (which tracks file upload progress), which is enabled by default in PHP.
By exploiting these vulnerabilities, attackers can remotely execute code on the server with web server process privileges without authentication.
According to PHP developers, all installations of unpatched versions that process web requests are at risk, including PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP, and LiteSpeed. The greatest risk lies with mod_php: researchers demonstrated an exploit for it in which an unauthenticated request triggers the attack even before the application code is executed.
What businesses should do
We’ve compiled key recommendations from the PHP developers and experts at Kaspersky.
Update PHP to versions 8.5.12, 8.4.27, 8.3.36, or 8.2.35. Start with systems accessible from the internet. The source code is available on the PHP website. If PHP is installed from distribution packages or provided by your hosting provider, check for updated packages: Linux distributions and LTS vendors received patches in advance. If the server is managed by a hosting provider, ask them to apply the update.
Versions 8.1 and older no longer receive updates from the PHP project. Check with your distribution vendor or LTS support provider to see if backported fixes are available, or switch to a supported branch. For PHP 7.4, 8.0, and 8.1, the developers have prepared minimal source code backports for manual compilation. These can also be found on the PHP website.
Restart PHP-FPM or the web server so that the processes load the new code, and verify the version using the `php -v` command. If PHP was installed from distribution packages, check the package changelog: the version number may remain the same.
Check the logs from October 20. A warning sign is when the web server process is performing actions it shouldn’t (running command shells, curl, wget, and the like). Kaspersky solutions help detect malicious activity in a timely manner. Attempts to exploit the described RCE vulnerabilities on Linux and Windows machines can be detected using Kaspersky NEXT EDR Expert, while exploitation of PHP2Shell in network traffic is detected by Kaspersky Anti Targeted Attack (KATA) with the NDR module.
If an immediate update is not possible
Only the immediate installation of patches will fully resolve the issue, but if that isn’t possible, it’s recommended to implement temporary measures. PHP developers provide step-by-step instructions for each vulnerability.
The more a company’s infrastructure grows, the more weak spots it has. And that’s true for all companies — from large corporations with deep pockets that are able to invest more in their defenses, to small and medium-sized businesses. And despite their more modest capabilities, smaller companies are just as tempting a target for attackers.
New times, new risks
Cloud platforms, remote access, and AI help businesses scale up and get more done. But they also give attackers extra ways to get into corporate infrastructure. For example, remote employees’ personal devices have become a foothold that’s hard to control. And thanks to AI, phishing emails are almost impossible to tell from real ones: users can no longer spot a scammer by their spelling mistakes.
There’s another reason why new threats keep appearing: companies are slowly learning how to handle the old ones. Take backups. Just a few years ago they were sold as a cure-all against ransomware. Now they’re less effective, because at some point attackers started stealing data and demanding payment not for decryption, but for not making it public. What’s at stake here isn’t just business processes, but the organization’s reputation too. We wrote about this scheme in more detail in a separate blog post.
The scariest part is that an attacked company’s problems don’t end on the day of the incident. The next few days (or even weeks) are often much harder. This is when the team needs to quickly deal with the aftermath of an attack and to make sure another can’t happen again. Growing companies often have to throw their entire small IT force at the problem, so getting back to normal work can take a long time.
Three Achilles’ heels at once
According to the Global Kaspersky B2B Market Pulse survey, most SMBs only think about security when circumstances force them to; for example, after an incident, when expanding their infrastructure, or when a potential partner asks them to strengthen their perimeter before signing a contract. Yet only a handful treat continuous cybersecurity monitoring as a core part of running their business.
When a company finally decides to get serious about security, it runs into three constraints: people, processes, and technology. Let’s look at each one.
People. In a company with up to 500 employees, the IT department is usually just a few people. They handle all technical issues and have little to do with cybersecurity. When things get tough, these specialists may simply lack the expertise to deal with things. Other employees can be a weak link too. Accountants, lawyers, and other staff with access to confidential information often aren’t familiar with security rules.
Processes.A small IT department always has a backlog with a long queue of tasks; therefore, fresh updates and patches don’t get installed right away. As a result, attackers have plenty of time to take advantage of a vulnerability that isn’t fixed in time. Meanwhile, other employees may not know how to respond to suspicious emails, or how to store and share passwords safely. And there’s simply no one to keep an eye on them.
Technology. Threats evolve alongside a business’s technology. For example, once a company has moved part of its infrastructure to the cloud, relying on a classic antivirus alone becomes pointless.
So small and medium-sized businesses become easy targets for attackers not so much due to their size, but because they don’t pay enough attention to the key building blocks of cybersecurity. As to what this can lead to, and how the tables can be turned in a company’s favor — read on for a real-world case that demonstrates this.
Case study: ERP system breach
Let’s look at a real-world attack on a company with 450 employees. When the incident happened, the company was migrating its ERP system to the cloud. The data had already been moved, but the protection hadn’t been set up yet. And that limbo state is exactly what let the attackers pull off their campaign.
The attackers used a fairly simple tactic. Posing as an external ERP integrator, they sent the finance department a file with an “adjusted invoice”. When an employee opened the attachment, the malicious code inside it exploited a vulnerability in a local application, and the attackers took control of the device. There they found saved credentials that let them log in to a support account and get administrator rights on the cloud ERP server. With that account, they got into the server and stole the company’s confidential data.
We analyzed the mistakes that made this incident possible, and came back to the same formula: people + processes + technology. The company did provide employee training, but it was mostly a formality and did nothing to help people learn to spot email phishing. Security updates weren’t installed right away either, which is why the computer had an unpatched vulnerability. On top of that, privileged credentials were stored right on the employee’s device, which gave the attackers a direct path to administrator access. Finally, the workstations had no EDR, and the cloud ERP server had no workload protection.
How SMBs can strengthen their defenses
The best way to start is with the three key areas we covered above. And this isn’t about simply throwing money at security. What’s needed is a clear, step-by-step plan that makes the most of the resources the company already has.
Hands-on training
Employees are a business’s first line of defense, so it’s especially important to help them form safe cyberhabits. Many companies understand this and duly run computer literacy training; however, often employees show up in droves, yet they keep on making the same old mistakes.
The problem is that most training isn’t relevant to the people taking it. People in the finance department know that attackers can send phishing emails, but what does that have to do with their job? To head off questions like that, we recommend combining cybersecurity theory with simulations of real cyberattacks. That way, employees can safely step into the shoes of an attacker’s target. Start by training employees who have privileged access to systems, and also those who handle confidential information.
To keep training from becoming a formality, track progress regularly. For example, send employees suspicious test emails from time to time, and measure how fast and how well they react: how many recipients click the link, and how many report the threat.
Structured processes
Set up clear priorities for installing updates and patches. This helps avoid two extremes: running for months with open gaps in the infrastructure, or rushing to fix every vulnerability without asking how critical it is for specific business processes. Setting up centralized vulnerability management also helps, which we explained how to do in a separate post. Don’t rush to expand the infrastructure or move it to the cloud until the company has proper protection and monitoring in place.
Another key point: assign an owner to every process. Someone should be personally responsible for regular cybersecurity training, installing updates, and the timing and methods of incident response. That way, security tasks get done on time — even when the IT team has limited resources.
Gradual rollout of security solutions
It’s impossible to build reliable business protection without specialized products, but buying everything at once won’t help either. New solutions should match the needs of the infrastructure. A good approach is to add tools as the need to do so comes up; for example, to prioritize vulnerabilities more effectively or to strengthen the security of email or cloud environments. One way to tailor security to the company’s needs is to use our Security Modules , which work with Kaspersky Next Optimum — a solution for growing SMB and mid-market businesses.
Another option is to move to a more advanced solution as the infrastructure grows. The foundation of cybersecurity can be laid with Kaspersky Next EDR Foundations, which provides basic endpoint protection and control. Then, when needed, the company can move up to solutions with a broader range of capabilities, from Kaspersky Next EDR Optimum to Kaspersky Next XDR Optimum, or straight to Kaspersky Next MXDR Optimum. Our most comprehensive offering — Kaspersky MXDR Optimum — combines advanced detection and response that goes beyond endpoints and adapts to the needs of a growing business, 24/7 monitoring, and proactive threat hunting and investigation by Kaspersky experts.
Microsoft’s October Patch Tuesday update stands out for two reasons. First, the total number of patches has reached four digits for the first time. Second, to help companies better understand the details of the vulnerabilities, Microsoft has decided to group them into risk cohorts. A cohort brings together vulnerabilities that are exploited under the same conditions and affect the same systems. They replace neither the severity rating, nor the individual bulletins for each CVE. While they’re unlikely to be suitable for comprehensive patch prioritization, they will help quickly assess which part of an infrastructure is currently most at risk.
Statistics corner
Of the 1,414 vulnerabilities addressed in the October Patch Tuesday release, 1,102 directly affect the Windows operating system, 134 — Microsoft Office, 36 — SharePoint Server, 22 — SQL Server, and seven — Exchange Server. The vast majority — 677 — lead to privilege escalation, 317 to remote code execution, 40 to security bypass, and 73 to denial of service (DoS).
There are 86 critical vulnerabilities on the list, of which 68 lead to RCE, 14 to privilege escalation, three to spoofing, and one to data leakage. In addition, two vulnerabilities were disclosed before being patched, and one is being exploited in real-world attacks.
CVE-2026-96951 — vulnerability in Windows Shell
The only zero-day vulnerability patched in the October update that attackers are already exploiting in real-world attacks is CVE-2026-96951 in Windows Shell (CVSS severity score: 8.8). Details on how it’s being exploited aren’t yet known; however, according to Microsoft’s description, it involves command injection that allows bypassing a security feature when accessing the system over a network. Microsoft credits experts at both Fujitsu and Anthropic for discovering the vulnerability, so it was likely found using neural networks.
CVE-2026-108398 and CVE-2026-92707: vulnerabilities disclosed before patches were released
Information about these two vulnerabilities was made public before the October update was released. In theory, this could mean that attackers attempted to exploit them ahead of time. These are:
CVE-2026-108398 — a vulnerability in the .NET Framework that allows bypassing WDAC protection;
CVE-2026-92707 — a vulnerability in the third-party MessagePack-CSharp serialization component, leading to a denial of service (DoS).
However, according to the security bulletins, neither has been observed in actual attacks yet.
What else to pay attention to
Several vulnerabilities patched by the latest update combine potentially serious consequences with relatively simple exploitation. Particular attention should be paid to three flaws that could potentially lead to arbitrary code execution:
Microsoft isn’t the only company that releases regular update packages for newly discovered vulnerabilities.
In its October update, Adobe fixed 174 defects in its products, including critical vulnerabilities (CVSS 10) in ColdFusion and Commerce/Magento. These systems are typically accessible from an external network, making them particularly attractive to attackers.
SAP’s October security bulletin lists 20 vulnerabilities, including CVE-2026-89111 — an SSRF vulnerability in the Cloud SDK for Python. This vulnerability, rated CVSS 10, allows attackers to steal OAuth tokens.
How to keep your company safe?
Recently, thanks to neural networks, the number of vulnerabilities detected and patched has been regularly breaking records. To avoid getting overwhelmed by the flood of patches, we recommend using an automated platform for tracking and assessing vulnerabilities . It helps not only identify and fix vulnerabilities, it also prioritizes them based on actual risks and automates the remediation process.
When it comes to protecting voice calls, security professionals typically focus on the security of the data carrying the audio — above all on encryption. Sooner or later though, the voice data stops being a stream of bits and becomes an analog signal that goes straight to the speakers. You might think that if someone is using headphones, nobody can listen in. But headphones are electronic devices, and their components have physical characteristics that can give away closely guarded secrets without any cyberattack at all. In a recent paper, researchers from two Hong Kong universities came up with InjectEave — an attack that exploits these properties to intercept the already-decoded analog audio signal.
How the InjectEave attack works
The electronic components in headphones convert data into analog form and amplify the signal to drive the speakers. This process generates weak electromagnetic emissions. Capturing those emissions and extracting information from them is a field unto itself, with many experts working in it. The main issue those experts face lies in the fact that stray emissions from headphones are too faint to intercept and analyze in practice.
The authors of InjectEave propose adding a radio transmitter that beams a signal at a specific frequency toward the target device. Those weak emissions from the headphones’ analog components modulate the transmitter’s signal. The altered signal can be intercepted from several meters away — provided the receiving antenna is sensitive enough. The attacker then analyzes the changes in the signal to reconstruct the original audio.
Technical details of the InjectEave attack
The main obstacle to intercepting electromagnetic emissions directly — without an external transmitter — is the relatively low signal frequency. For example, the audio amplifier in headphones operates at frequencies from 20Hz to 20kHz. By contrast, electromagnetic emissions are easiest to intercept at frequencies measured in megahertz. The researchers were able to find a transmitter frequency that would also “amplify” the low-frequency signal of interest.
The researchers ran a series of experiments. They used antennas to transmit the “amplifying” signal and receive the result, and a spectrum analyzer to capture and process the data. The authors have posted the raw recordings of the intercepted signals on a dedicated microsite, so anyone can check how well the attack works. In most cases, getting a usable result didn’t even require complex processing or recognition. Conversations picked up from wireless headphones are clearly intelligible despite the elevated noise level.
In total, the authors tested three models of wireless headphones and two models of basic wired headsets. The effective eavesdropping distance varied from one to six meters depending on the model. With a wired telephone, they were able to demonstrate two-way interception. They could hear not only the sound from the speakers but also the potential victim’s speech, which the headset’s built-in microphone and analog-to-digital converter had captured.
Advantages and limitations of the method
Any attack that requires getting very close to the target is challenging by definition. The need to tune the frequency of the “amplifying” radio signal for each specific device makes the attack even harder. In their experiments, the researchers had to use frequencies from one to eight megahertz.
Despite these caveats, the main achievement of this attack is its being able to target the quirks of the analog components in an electronic device. These components inherently produce stray emissions, and there’s almost no way to reduce them.
Similar surveillance methods that intercept a digital signal can be stopped by adding a layer of encryption. When it comes to the analog side of a device, protecting the signal itself is much harder. This research is another reminder of how complex and unpredictable even very simple, familiar devices can be. Sometimes even the strongest protection for digital data is powerless once that information is converted to analog form.
Login
