I covered a fascinating post by Alexander Moch at ERNW about a boot-level Linux vulnerability that lets attackers inject code from the initramfs debug shell. Even with Secure Boot and encryption, a few key presses can drop you to a shell and allow persistent malware to be added.
Luckily, the fix is simple and involves kernel parameters. I break it down here:
https://nerds.xyz/2025/07/linux-initramfs-security-flaw-secure-boot-bypass/
Curious what others are doing to harden this layer.
|Reference: [Ongoing Campaign Abuses Microsoft 365’s Direct Send to Deliver Phishing Emails](https://www.varonis.com/blog/direct-send-exploit)|
|:-|
**Key Points:**
* **Phishing Campaign:** Varonis' MDDR Forensics team uncovered a phishing campaign exploiting Microsoft 365's Direct Send feature.
* **Direct Send Feature:** Allows internal devices to send emails without authentication, which attackers abuse to spoof internal users.
* **Detection:** Look for external IPs in message headers, failures in SPF, DKIM, or DMARC, and unusual email behaviors.
* **Prevention:** Enable "Reject Direct Send," implement strict DMARC policies, and educate users on risks.
For technical details, please see more in reference (above).
Could anyone share samples or real-world experiences about this *(for education and security monitoring)*?
RAWPA helps security researchers and penetration testers with hierarchical methodologies for testing.
**This is not a "get bugs quick scheme"**. I fully encourage manual scouring through JS files and playing around in burp, RAWPA is just like a guided to rejuvenate your thinking.
Interested ? Join the testers now
[https://forms.gle/guLyrwLWWjQW61BK9](https://forms.gle/guLyrwLWWjQW61BK9)
Read more about RAWPA on my blog: [https://kuwguap.github.io/](https://kuwguap.github.io/)
#
**\[Disclosure: I work at CyberArk and was involved in this research\]**
We've completed a security evaluation of the Model Context Protocol and discovered several concerning attack patterns relevant to ML practitioners integrating external tools with LLMs.
**Background:** MCP standardizes how AI applications access external resources - essentially creating a plugin ecosystem for LLMs. While this enables powerful agentic behaviors, it introduces novel security considerations.
**Technical
I wrote a blog [post](https://asherfalcon.com/blog/posts/2) discussing how I hid images inside DNS records, you can check out the web viewer at [https://dnsimg.asherfalcon.com](https://dnsimg.asherfalcon.com) with some domains I already added images to like [asherfalcon.com](http://asherfalcon.com) and [containerback.com](http://containerback.com)
This issue affects systems where KTelnetService and a vulnerable version of Konsole are installed but at least one of the programs telnet, rlogin or ssh is not installed. The vulnerability is in KDE's terminal emulator Konsole. As stated in the [advisory by KDE](https://kde.org/info/security/advisory-20250609-1.txt), Konsole versions < 25.04.2 are vulnerable.
On vulnerable systems remote code execution from a visited website is possible if the user allows loading of certain URL schemes (telnet:
Hi all, I discovered suspicious behavior and possible malware in a file related to the official MicroDicom Viewer installer. I’ve documented everything including hashes, scan results, and my analysis in this public GitHub repository:
[https://github.com/darnas11/MicroDicom-Incident-Report](https://github.com/darnas11/MicroDicom-Incident-Report)
Feedback and insights are very welcome!