Security news weekly round-up - 3rd April 2026

#security

Malware, vulnerability, and research in computer security are mostly what we'll talk about in this week's security review. As always, you should know the threat out there and you're responsible for acting accordingly.

As always, my name is Habdul Hazeez. Welcome to this week's review.

Fake VS Code alerts on GitHub spread malware to developers

If you're a developer who receives lots of email notifications from GitHub, be careful of the one that you respond to.

Here is what's going on:

The discussions are posted in an automated way from newly created or low-activity accounts across thousands of repositories within a few minutes, and trigger email notifications to a large number of tagged users and followers.

The posts include links to supposedly patched versions of the impacted VS Code extensions, hosted on external services such as Google Drive.

Although Google Drive is obviously not the official software distribution channel for a VS Code extension, it’s a trusted service, and users acting in haste may miss the red flag.

Stolen Logins Are Fueling Everything From Ransomware to Nation-State Cyberattacks

Username is correct. Password is correct. Successful login. Now, the question you should ask: was it a legitimate user that just accessed your system, or, was it an imposter? If you have the chance to ask yourself that question. Good for you. If an alert goes off afterwards, know that something could be wrong, e.g., a ransomware attack.

From the article:

The theft and resale of credentials operates on an industrial scale. Fueled by the rise of increasingly more sophisticated infostealers, stolen credentials are packaged into ‘logs’ and sold to criminals on the black market.

Ransomware has been one of the primary beneficiaries of stolen credentials. More than 7,000 incidents and 129 active groups were tracked through 2025. At the same time, ransom payments decreased slightly from $892M in 2024 to $820M in 2025.

Digital assets after death: Managing risks to your loved one’s digital estate

From me to you: I don't wish that you die any moment from now. Nonetheless, please, start having that conversation with your loved ones now. What happens to your digital life when you are no more? How will your family get hold of your digital assets? And questions like that. That's what the article is trying to raise awareness about. The message is for me and you.

If you still need more convincing, here is an excerpt from the article that should do the trick:

It’s important to understand that, while most big tech companies offer the ability to transfer access to a “legacy contact,” if you don’t take advantage of this before passing on, the chances are that no one will be able to access your accounts

Microsoft Warns of WhatsApp-Delivered VBS Malware Hijacking Windows via UAC Bypass

By reading the article, it's evident that if attackers successfully infect a system with this VBS malware, they can exfiltrate users' private data or deploy more malware. Also, by reading the article's title, you can ask: Who would knowingly execute a VBS malware via WhatsApp? The answer: the attackers use social engineering to get the user to do it.

Here is what's going on:

The activity begins with the attackers distributing malicious VBS files via WhatsApp messages that, when executed, create hidden folders in "C:\ProgramData" and drop renamed versions of legitimate Windows utilities like "curl.exe" (renamed as "netapi.dll") and "bitsadmin.exe" (renamed as "sc.exe").

Upon gaining an initial foothold, the attackers aim to establish persistence and escalate privileges, ultimately installing malicious MSI packages on victim systems. This is achieved by downloading auxiliary VBS files hosted on AWS S3, Tencent Cloud, and Backblaze B2 using the renamed binaries.

New Rowhammer attacks give complete control of machines running Nvidia GPUs

These types have of attack have come to public light since 2014. It's, 2026 and researchers are still discovering different forms of the attack. This time, in the form of GDDRHammer and GeForge.

Here is what's going on:

GDDRHammer can manipulate the memory allocator to break isolation of GPU page tables—which, like CPU page tables, are the data structures used to store mappings between virtual addresses and physical DRAM addresses—and user data stored on the GPU. The result is that the attacker acquires the ability to both read and write to GPU memory.

GeForge, too, uses novel hammering patterns and memory massaging to corrupt GPU page table mappings in GDDR6 memory to acquire read and write access to the GPU memory space. From there, it acquires the same privileges over host CPU memory.

'NoVoice' Android malware on Google Play infected 2.3 million devices

If you have a device that has security updates from 2021 to the current date, you can mitigate the flaws targeted by this NoVoice malware. Otherwise, you are vulnerable.

Here is what NoVoice can do to an infected device:

According to McAfee researchers, the threat actor concealed malicious components in the com.facebook.utils package, mixing them with the legitimate Facebook SDK classes.

An encrypted payload (enc.apk) hidden inside a PNG image file using steganography is extracted (h.apk) and loaded in system memory while wiping all intermediate files to eliminate traces.

The malware then contacts the command-and-control (C2) server and collects device information such as hardware details, kernel version, Android version (and patch level), installed apps, and root status, to determine the exploit strategy.

Critical Vulnerability in Claude Code Emerges Days After Source Leak

It's one thing to accidentally leak the source code of one of your tools and it's another for someone to discover a vulnerability. Now, we can say you have to do two things: build another version of the tools whose code got leaked, and ensure that it does not get to the public. Second, find a way to protect the users of the current version from the vulnerability. Anyways, we can say, Anthropic has a lot to deal with.

From the article, here is some brief information about the vulnerability:

The problem stems from Anthropic’s desire for improved performance following the discovery of a performance issue: complex compound commands caused the UI to freeze. Anthropic fixed this by capping analysis at 50 subcommands, with a fall back to a generic ‘ask’ prompt for anything else. The code comment states, “Fifty is generous: legitimate user commands don’t split that wide. Above the cap we fall back to ‘ask’ (safe default — we can’t prove safety, so we prompt).”

The flaw discovered by Adversa is that this process can be manipulated. Anthropic’s assumption doesn’t account for AI-generated commands from prompt injection — where a malicious CLAUDE.md file instructs the AI to generate a 50+ subcommand pipeline that looks like a legitimate build process.