Are You Down With RDP? – PSW #873

Paul Asadoorian

1. SysOwned, Your Friendly Support Ticket – SysAid On-Premise Pre-Auth RCE Chain (CVE-2025-2775 And Friends)
2. SonicBoom, From Stolen Tokens to Remote Shells – SonicWall SMA (CVE-2023-44221, CVE-2024-38475)
3. Decrypting Firmware: Uncovering Hidden Threats Through Binary Reverse Engineering with Hopper
4. SAP NetWeaver CVE-2025-31324 Exploitation
5. TheWizards APT group uses SLAAC spoofing to perform adversary-in-the-middle attacks
6. Linux Kernel Exploitation: CVE-2025-21756
7. Supercharging Ghidra: Using Local LLMs with GhidraMCP via Ollama and OpenWeb-UI
8. Hello 0-Days, My Old Friend: A 2024 Zero-Day Exploitation Analysis
9. Multiple Vulnerabilities in HP Wolf Security Controller / HP Sure Access Enterprise / HP Sure Click Enterprise
10. io_uring Rootkit Bypasses Linux Security Tools – ARMO
11. DSA-2025-150: Security Update for Dell SupportAssist OS Recovery for a Third-Party Component Driver Vulnerability
12. IMPORTANT! Paragon Driver Security Patch for All Products of Hard Disk Manager Product Line (Biontdrv.sys)
13. iOS and Android juice jacking defenses have been trivial to bypass for years
14. Jumping the line: How MCP servers can attack you before you ever use them
15. SSD Advisory – How MiraclePtr Crushed Two Sandbox Escapes – SSD Secure Disclosure
16. AES & ChaCha — A Case for Simplicity in Cryptography
17. Glitching STM32 Read Out Protection – Anvil Secure
18. This Week In Security: XRP Poisoned, MCP Bypassed, And More
19. Security Advisory: Remote Code Execution on Viasat Modems (CVE-2024-6198)
20. How MCP servers can steal your conversation history
21. Fire In The Hole, We’re Breaching The Vault – Commvault Remote Code Execution (CVE-2025-34028)
22. IngressNightmare: Understanding CVE‑2025‑1974 in Kubernetes Ingress-NGINX
23. SonicWall Sonicos Versions 7.1.x and 8.0.x
24. How to Root Android Phones – Black Hills Information Security, Inc.

    Great tutorial to get you started on Android rooting. I highly recommend that you play around with Android devices and try to root them. It's fun and sets the stage for you to install other Android operating systems (like Graphene) and Kali NetHunter. Keep this post bookmarked!

25. CVE-2025-23016 – Exploiting the FastCGI library

    This is a great vulnerability and exploit. I suspect there are many IoT and embedded systems that still have this vulnerability, as FastCGI is pretty common on these platforms. Some details:

    • Root Cause: An integer overflow in the ReadParams function when processing HTTP parameters (nameLen and valueLen). On 32-bit systems, excessively large values for these parameters cause a miscalculation in memory allocation, leading to a heap buffer overflow.
    • Impact: Attackers can overwrite heap memory to execute arbitrary code, often gaining full control of the device. This is especially dangerous for embedded systems lacking modern protections like ASLR or NX
    • You have to either use UNIX sockets (rather than TCP) for FastCGI communications and/or upgrade to the latest version of FastCGI
26. SSD Advisory – Samsung MagicINFO Unauthenticated RCE – SSD Secure Disclosure

    MagicINFO is software from Samsung that provides digital signage: "Samsung MagicINFO is an all-in-one digital signage solution designed to manage, schedule, and monitor multimedia content across Samsung commercial displays. It provides a centralized platform for content creation, device management, and data analysis, supporting a wide range of business applications such as retail, transportation, hospitality, healthcare, and corporate environments" - There is a really easy to exploit vulnerability, already observed in the wild, already in Mirai!

27. Hello my perverted friend – The Hacker Factor Blog

    I think this is spray and pray (forgive the pun), they send out LOTS of emails which is cheap, and get 10 people to pay (According to Bitcoin wallet analysis from the comments), then it's a win for the attackers. Good to educate folks about these scams.

28. Cybersecurity Firm CEO Arrested for Planting Malware in Hospital Systems

    "Jeffrey Bowie, the CEO of a local cybersecurity firm, has been arrested for allegedly planting malware on computers at SSM St. Anthony Hospital. Bowie, who until recently touted himself as a leader in protecting businesses from cyber threats, now faces charges that he became the very threat he promised to prevent. Police say the incident unfolded on August 6, 2024, when an alert employee noticed Bowie, a non-hospital staff member, using a computer designated strictly for employees. The employee’s quick-thinking report set off an immediate internal investigation by the hospital’s IT teams." - Speculation, but obviously they are looking at ties between the healthcare provider and the cybersecurity firm to see if the malware was put there to drum up business, which would be awful. We used to speculate that the anti-virus companies made the viruses, but there has never been evidence of this (at least that I know of). Also, good for the employee who spotted this behavior, had they not reported it, who knows how long the malware could have lingered.

29. Remote Code Execution in ZYXEL FLEX-H Series

    Really neat vulnerability and exploit. Using Postgres over an SSH port forward the researcher was able to read files and execute commands using the "COPY from" functionality in Postgres. More research on this comes from Marco as well: https://security.humanativaspa.it/local-privilege-escalation-on-zyxel-usg-flex-h-series-cve-2025-1731/ (Marco just did one of the MS podcasts too).

30. Windows RDP lets you log in using revoked passwords. Microsoft is OK with that.

    This is crazy that Microsoft won't fix this: "The ability to use a revoked password to log in through RDP occurs when a Windows machine that’s signed in with a Microsoft or Azure account is configured to enable remote desktop access. In that case, users can log in over RDP with a dedicated password that’s validated against a locally stored credential. Alternatively, users can log in using the credentials for the online account that was used to sign in to the machine. Even after users change their account password, however, it remains valid for RDP logins indefinitely. In some cases, Wade reported, multiple older passwords will work while newer ones won’t. The result: persistent RDP access that bypasses cloud verification, multifactor authentication, and Conditional Access policies." To make matters worse, here is the breakdown from independent security researcher Daniel Wade who made the discovery and disclosed it to Microsoft:

    • Old credentials continue working for RDP—even from brand-new machines.
    • Defender, Entra ID, and Azure don’t raise any flags.
    • There is no clear way for end-users to detect or fix the issue.
    • No Microsoft documentation or guidance addresses this scenario directly.
    • Even newer passwords may be ignored while older ones continue to function.
31. AirBorne: Wormable Zero-Click RCE in Apple AirPlay Puts Billions of Devices at Risk

    This is a sweet discovery: "Oligo has demonstrated that two of the vulnerabilities (CVE-2025-24252 and CVE-2025-24132) allow attackers to weaponize wormable zero-click RCE exploits. This means that an attacker can take over certain AirPlay-enabled devices and do things like deploy malware that spreads to devices on any local network the infected device connects to. This could lead to the delivery of other sophisticated attacks related to espionage, ransomware, supply-chain attacks, and more." These vulnerabilities are very widespread:

    • While not every Apple device worldwide is vulnerable to RCE via AirBorne, Apple stated in * January 2025 that there are 2.35 billion active Apple devices across the globe. *In 2018 Apple indicated that there were over 100 million active MacOs users globally.
      *Other Apple devices like iPhones, Apple TV, vision pro are affected by different AirBorne vulnerabilities. Devices like iPhones require users to enable the AirPlay receiver in the phone settings.
    • Additionally, the number of third-party audio devices that support AirPlay can be estimated in the 10s of millions.
    • While the exact number of CarPlay-enabled devices is not known, it is widely-used and available in over 800 vehicle models.

    It's not just Apple devices, but any speaker or car that uses the AirPlay SDK. This is most concerning for me because they may not get updates, ever. I was just looking at these devices for an older vehicle to enable Carplay and Android Auto, many Chinese manufactured devices exist, likely running Android, and likely never will be updated, and likely already have backdoors and malware!

32. Vibe Hacking: Finding Auth Bypass and RCE in Open Game Panel

    I loved this write-up, very honest and not over-hyping "Vibe coding" or "Vibe hacking" in any way: "Ultimately, what I found from this mini project is that full blown vibe based security research isn't quite there yet. The vulnerabilities we uncovered were pretty straightforward, but it still took some interpretation and manual digging to turn ideas into working exploits. What I am convinced of though is that even today, tools like Cursor are great for exploring large and unfamiliar codebases." - I do like the approach, tools like Cursor are just that, a tool for security researchers that hopefully make our lives easier. But we still need to do the final mile (which means we still need to understand software vulnerabilities and exploits).

33. How We Discovered Planet Technology Network Device Vulnerabilities

    Great research! CISA released an advisory about these devices, industrial switches. There were not details so the researcher reversed the hardware and firmware. Spoiler: "I was successful in identifying the vulnerabilities from the CISA report. But I also spotted a couple of other vulnerabilities, not just in these switches (and likely other models), but also in the network management tool that’s used to remotely manage whole fleets of Planet devices in organizations." - Other things I learned: check out the new version of binwalk and using Python to automate some of the analysis looked awesome. The Python script (Available here: ) does the following:

    • Scans a given directory for any ELF binaries
    • Creates a Ghidra project
    • Loads each ELF file and runs Ghidra’s default analysis
    • Searches for a set of functions and xrefs
    • If a function and xref is found, the decompiled code is written to a directory

    Love this so much.

34. ESP32-DIV: Your Swiss Army Knife for Wireless Networks

    More stuff that I want to build!

35. How a Single Line Of Code Could Brick Your iPhone

    More sweet research: *"To summarize, Darwin notifications:

    • Require no special privileges for receiving
    • Require no special privileges for sending
    • Are available as public API
    • Have no mechanism for verifying the sender

    Considering these properties, I began to wonder if there were places on iOS using Darwin notifications for powerful operations that could potentially be exploited as a denial-of-service attack from within a sandboxed app. You’re reading this blog post, so I’ve already spoiled it: the answer was “yes”."*

36. Kali Linux warns of update failures after losing repo signing key

    You need to update your keyring: "sudo wget https://archive.kali.org/archive-keyring.gpg -O /usr/share/keyrings/kali-archive-keyring.gpg" - I don't keep every Kali instance updated as I use them as a utility, so this is good to know. Also, key management, or mis-management, happens.

37. Remote Desktop Puzzle Attack Let Hackers Exfiltrate Sensitive Data From Organization

    This is awesome: "In a recent case study, Pen Test Partners investigated a data breach where an attacker had deliberately wiped traditional evidence including Windows Event Logs, TerminalServices logs, and Security event logs. Despite these anti-forensic measures, investigators discovered the RDP bitmap cache folder remained intact. Using specialized tools including BMC-Tools and RdpCacheStitcher, investigators extracted and reconstructed over 8,000 bitmap cache files from the compromised system." - While this is not new, I love how they used it in a forensics investigation.

Bill Swearingen

1. Stop the Spreadsheet Madness: Visualize Your Atomic Red Team Tests with VECTR

   OMG I love articles like this. This post starts off talking about purple teaming tools, but ends up dropping real trade-craft on how to threat hunt your environment. If you are on the blue team, read this article!

2. Millions of Apple Airplay-enabled devices can be hacked via Wi-Fi

   On Tuesday, researchers from the cybersecurity firm Oligo revealed what they’re calling AirBorne, a collection of vulnerabilities affecting AirPlay, Apple’s proprietary radio-based protocol for local wireless communication. Bugs in Apple’s AirPlay software development kit (SDK) for third-party devices would allow hackers to hijack gadgets like speakers, receivers, set-top boxes, or smart TVs if they’re on the same Wi-Fi network as the hacker’s machine.

3. How are cyber criminals rolling in 2025?

   uh. Yikes. This site lists some google dorks that show .Gov and .Edu sites that have been hacked and used to host pirated movies. Who needs to pay for cloud hosting!?

4. TeleMessage, a modified Signal clone used by US government officials, has been hacked

   A hacker has exploited a vulnerability in TeleMessage, which provides modded versions of encrypted messaging apps such as Signal, Telegram, and WhatsApp, to extract archived messages and other data relating to U.S. government officials and companies who used the tool

Jeff Man

1. Lt. Governor Miller and Secretary Coker Bring Maryland’s Cyber Prominence to Center Stage at RSA 2025

   I was able to attend the breakfast sponsored by the Maryland Dept. of Commerce last week at RSAC Conference where Lt. Gov. Miller and Secretary Coker participated in a fireside chat talking up the merits of bringing businesses to the state of Maryland. Hey - I'm born and raised in Maryland, I have to show some respect!

2. The Death and Rebirth of Vulnerability Management – HD Moore

   I had a fascinating discussion with HD Moore last week at RSAC Conference, and I'd love to get him on the show sometime to continue the discussion of vulnerabilities, detection, management, prioritization, not to mention CVEs, CVSS scores, disclosure, patching, the whole nine yards.

3. Arctic Wolf Observes Exploitation of Path Traversal Vulnerability in Samsung MagicINFO 9 Server (CVE-2024-7399)

   High severity vulnerability made public in August 2024. No malicious activity detected. Proof of concept exploit published in Aprik 2025. Exploitation detected within days. Hrmmmmm...

4. A single burnt wire caused horrifying 90-second Newark airport air traffic control outage

   Maybe this isn't a cyber attack, but it sure seems like we should talk about it.

5. OCC Cyber Breach: Undetected for 8 Months, Exposing Sensitive Data

   "...a single compromised account can really jeopardize an entire organization's security."

   You think???

6. Cyber breach reverberates at Nova Scotia Power more than a week later

   Nova Scotia Power is remaining tight-lipped about the details of a cyber breach that has forced the company to pause billing unauthorized access into parts of its network and servers The company has noted major billing and customer service issues the personal information of some customers has been taken Three words.... P... C... I....

7. 5.5 million patients’ information exposed by major healthcare data breach

   Connecticut based Yale New Haven Health, has revealed that a data breach affected more than 5.5 million people. The information leaked included patient names, dates of birth, postal and email addresses, phone numbers and more. With the exception of the "and more" this isn't a HIPAA issue - which is the only thing most healthcare systems worry about from a cybersecurity perspective.

   The article goes on to say, "some healthcare-related data" was compromised. Still sounding like a stretch...no HIPAA violation???

Joshua Marpet

1. Broadcom is…not going to be popular. VMWare Cease and Desist

   VMware/Broadcom is sending out cease and desist letters to vmware perpetual license holders. Audit rights and no patching. What a great security company!!!

2. Culture change is hard!

Larry Pesce

1. Proposing a Software Defined Radio based “AI Battle Buddy”
2. GlobalX, Airline for Trump’s Deportations, Hacked
3. The Signal Clone the Trump Admin Uses Was Hacked
4. Hundreds of e-commerce sites hacked in supply-chain attack
5. Airborne: Wormable Zero-Click RCE in Apple AirPlay Puts Billions of Devices at Risk
6. Open source project curl is sick of users submitting “AI slop” vulnerabilities

Lee Neely

1. Pushing passkeys forward: Microsoft’s latest updates for simpler, safer sign-ins

   One year after year Microsoft began offering passkey support for consumer accounts, the company has announced that all new Microsoft accounts will be passwordless by default. The move is intended to protect customers’ credentials from stuffing, brute force, and phishing attacks. Current Microsoft customers who have not yet adopted passkeys will be encouraged to do so when they sign into their accounts. Other companies, including Apple and Google, are also developing passkey support under the aegis of the FIDO Alliance.

   The good news is you will be using passkeys with your Microsoft account and you can start setting passkeys on your Microsoft accounts today, and they will default to the strongest authentication option available. The bad news is you have to use Microsoft Authenticator, as Google Authenticator, Authy and similar apps are incompatible with their system, if you want to ditch your reusable password, which we should do, so authenication can't fall back to this option.

   https://arstechnica.com/security/2025/05/microsoft-pushes-unphishable-logins-forward-with-new-sign-in-options/

2. Backdoor found in popular ecommerce components

   Researchers from Sansec have discovered a supply chain attack that targeted multiple vendors of e-commerce software. In all Sansec found 21 Magento extensions infected with the same malware; Sansec estimates that the malware has infected between 500 and 1,000 online merchants. While the software was infected six years ago, it was activated only last month. The malware executes in site visitors’ browsers, and it capable of stealing payment card data and other sensitive information. Sansec recommends that all e-commerce sites running software from the affected vendors – Tigren, Magesolution (MGS), and Meetanshi, all of which are based on Magento.

   This is tricky as you're looking at an ecommerce package bundled on top of Adobe Commerce/Magento to include the card skimming malware, so you're dependent on the provider of the package, not Adobe, to provide the update, and the vendors have not yet announced updated versions. You need to check for the IOCs  in the Sansec report, (the backdoor code is the same, but checksum, path and license filename is unique for each vendor) and if found, remove it.

3. Spain and Portugal Power Outages Spark a Surge in Phishing

   While the massive power outage on the Iberian peninsula on April 28 may not have been caused by a cyberattack but rather by an atmospheric phenomenon, the resulting widespread disruption was opportunistically leveraged as part of a phishing campaign targeting speakers of Portuguese and Spanish. Cofense has published a blog post showing email messages sent during the blackout impersonating the Portuguese national airline TAP Air Portugal, invoking an EU regulation on air passengers' rights and offering compensation for delayed or cancelled flights through a WordPress phishing page aimed at collecting the target's name, date of birth, mailing address, email address, phone number, and credit card details. Teaching family and users to step back and verify communication is genuine, particularly when emotions are running hot, is difficult and essential. To include double checking the reference used to get a legitimate POC because typos happen.

4. Man Admits Hacking Disney and Leaking Data Disguised as Hacktivist

   A 25-year-old man from California has pleaded guilty to hacking Disney systems and leaking data under the guise of a hacktivist collective, the Justice Department announced. According to the DoJ, Ryan Mitchell Kramer has pleaded guilty to accessing a computer and obtaining information, and threatening to damage a protected computer, as well as to two felony charges that each carry a prison sentence of up to five years. Kramer posted a program online that claimed to be an AI image generator. In fact, the program contained malware that gave Kramer access to the computers of people who downloaded it.

   The malware was downloaded on personal computers and accessed stored credentials for the employee's work-related slack channel. Tools like Slack/Discord/etc. are the norm in modern team communication and collaboration. Two things to follow up on here. First, make sure that work related services, chat/email/file shares, require MFA, second, review your access control model, considering the type of data available in these services.

5. Hackers launch ‘serious’ attacks against Georgia school district, New Mexico university

   Multiple US School districts have been experiencing cyber attacks in the past few weeks. Often falling back to chrome books/etc. until school desktops/networks are verified clean.

   The timing is unfortunate given the proximity of the end of the term/graduation. Parents and students should make sure you're aware of access methods for your campus/schoolwork, so you know what plan B is. As an educator, beyond ensuring you have cyber hygiene and incident response capabilities, include a frank discussion about operating, including what can be offline, during an incident.

6. Raytheon settles its civil case over federal security regs, while feds indict Ukrainian national for alleged Nefilim ransomware activities.

   US government contractors Raytheon Company, RTX Corporation, Nightwing Group LLC, and Nightwing Intelligence Solutions LLC will collectively pay $8.4 million for violations of the False Claims Act. According to the Department of Justice, "the settlement resolves allegations that Raytheon and its then-subsidiary Raytheon Cyber Solutions, Inc. (RCSI), failed to implement required cybersecurity controls on an internal development system that was used to perform unclassified work on certain DoD contracts. The United States alleged that Raytheon and RCSI failed to develop and implement a system security plan for the system, as required by DoD cybersecurity regulations, and failed to ensure that the system complied with other cybersecurity requirements … ."

   Have you ever wondered what happens if you don't create a security plan and implement controls for DoD data, as required by 800-171/800-53, let alone claim to do so when you haven't? 800-171 applies to contractor owned systems and allows self-attestation, while 800-53 to government owned systems and requires asuthority to operate (ATO) from a federal authorizing official.  CMMC 2.0 is intended to discover and address gaps such as this. The false claim came from Raytheon's former director of engineering, as a whistleblower, who will receive more than $1.5 million of the settlement. https://www.justice.gov/opa/pr/raytheon-companies-and-nightwing-group-pay-84m-resolve-false-claims-act-allegations-relating

7. High-Severity Vulnerabilities in MicroDicom DICOM Viewer

   The CISA has published an ICS Medical Advisory warning of two high-severity vulnerabilities affecting MicroDicom DICOM Viewer. Digital Imaging and Communications in Medicine (DICOM) is the standard for sharing medical imaging information and related data. The viewer is used to both view images as well as burn them onto CD/DVDs which can be viewed without the software. The flaws, CVE-2025-36521, out of bounds write , CVSS score 8.8, and CVE-2025-36251, out of bounds read, CVSS score 8.8, require user interaction to open a specially crafted .DCM file. The fix is to update to 2025.2, then make sure that your DICOM viewer is not Internet facing, use a VPN if remote access to the viewer is needed. Also, use caution opening DCM files, making sure they are genuine.

8. Airborne: Wormable Zero-Click RCE in Apple AirPlay Puts Billions of Devices at Risk

   Oligo Security has published a blog post describing their discovery of 23 vulnerabilities in the Apple AirPlay software development kit (SDK) and AirPlay Protocol.

   Apple released fixes for AirBorne in iOS/iPadOS 18.4, macOS 13.7.5, 14.7.5 and 14.5, visionOS 2.4 and watchOS 11.4.  Make sure that your devices are running these versions or newer. Setting AirPlay to "Current User" restricts access to devices logged into the same Apple Account. If you're not using AirPlay receiver, turn it off. Make sure your IoT devices run current firmware, and consider monitoring/limiting use of AirPlay (port 7000) on your network, document this decision so you don't have a recurring agenda item.

9. British Library avoids investigation over ransomware attack, praised again for response

   The UK Information Commissioner’s Office (ICO) will not investigate a 2023 ransomware attack that disrupted the British Library’s operations for months. The Library reported the incident to ICO in October 2023 and subsequently published a detailed review of the incident in March 2024: the “likely … point of entry” was determined to be a server that was not protected with multi-factor authentication (MFA). In a recent statement, the ICO writes that "further investigation would not be the most effective use of [their] resources."

   Kudos to IOC for their work investigating the incident, identifying the root cause and subsequent remediation. When the regulator says your investigation and follow-up actions are sufficient, take that as a win. Beyond this being a clue for your incident response plans, make sure you're looking for issues like inconsistent application of MFA before the attackers find it for you. Make sure remaining risks are both documented and accepted by someone authorized to do so.

10. Windows RDP lets you log in using revoked passwords. Microsoft is OK with that.

    Microsoft informed Daniel Wade, an independent security researcher, that he is not the first to express alarm that Windows Remote Desktop Protocol (RDP) allows logins using revoked passwords, an issue which Wade says "isn't just a bug. It's a Trust Breakdown" because users expect a password change to curb unauthorized access. Wade's report states that "even newer passwords may be ignored while older ones continue to function," bypassing cloud verification, MFA, and Conditional Access policies to allow persistent RDP access. Microsoft has been aware of the issue since 2023 and does not consider it a vulnerability, claiming it is "a design decision to ensure that at least one user account always has the ability to log in no matter how long a system has been offline."

    It's been a bit since I've had the "It's a feature not a bug" conversation. Given the response from Microsoft, this isn't behavior which is likely to change and should be understood as part of the risks for using RDP.  Microsoft updated their Windows Login Scenarios article (link below) with a big red box explaining this behavior. This cache behavior applies to local logins, which include RDP. You'll want to read that article a couple of times to understand how the login scenarios differ. Make sure that you're following current security guidelines for protecting RDP services, to include not exposing them to the Internet, as well as considering alternative services for remote access/support. https://learn.microsoft.com/en-us/windows-server/security/windows-authentication/windows-logon-scenarios

Sam Bowne

1. Windows RDP lets you log in using revoked passwords. Microsoft is OK with that.

   The problem is authenticating with a cloud account, the way Microsoft almost forces you to in Windows 11. If you change your cloud password, the old password still works for RDP. Microsoft, incredibly, regards this as intended behavior and has no intention of fixing it.

2. H 540: RDP and Old Passwords

   I wrote this up as a hands-on project for my students. An old hotmail password works for RDP. Also, I cannot find any way to remove the cached passwords from Windows 11. Removing Cached Logons in Group Policy does nothing, and clearing cached credentials in Credential Manager also does nothing. I think it is insane for Microsoft to ignore this vulnerability. Someone, perhaps a third-party researcher, needs to make a tool to clear out the old credentials.

3. Altman’s eyeball-scanning biometric blockchain orbs officially come to America

   World, previously Worldcoin, allows you to verify you are an actual person so that you can log into services that require the platform. It works with biometric scanning and uses a blockchain, and a cryptocurrency. Regulators around the world have raised concerns about this whole idea of an upstart collecting and storing people's biometric data, but it's coming to America now anyway.

4. North Korean operatives have infiltrated hundreds of Fortune 500 companies

   Security leaders at Mandiant and Google Cloud say nearly every major company has hired or received applications from North Korean nationals working on behalf of the country’s regime. “Nearly every CISO that I’ve spoken to about the North Korean IT worker problem has admitted they’ve hired at least one North Korean IT worker, if not a dozen or a few dozen.”

5. A DOGE recruiter is staffing a project to deploy AI agents across the US government

   Anthony Jancso, cofounder of AcclerateX, a government tech startup, is looking for technologists to work on a project that aims to have artificial intelligence perform tasks that are currently the responsibility of tens of thousands of federal workers.

   The company posted: “Outdated tech is dragging down the US Government. Legacy vendors sell broken systems at increasingly steep prices. This hurts every American citizen.”

   A critic said: “We want our government to be something that we can rely on, as opposed to something that is on the absolute bleeding edge. We don't need it to be bureaucratic and slow, but if corporations haven't adopted this yet, is the government really where we want to be experimenting with the cutting edge AI?”

6. Messaging app seen in use by Mike Waltz suspends service after hackers claim breach

   It was hacked twice, by different actors. It was apparently sending messages unencrypted to some insecure storage service.

7. TM SGNL, the obscure unofficial Signal app Mike Waltz uses to text with Trump officials

   Micah Lee analyzes the app, finding its source (Israel) and how it works.

8. New “Bring Your Own Installer” EDR bypass used in ransomware attack

   A new "Bring Your Own Installer" EDR bypass technique is exploited in attacks to bypass SentinelOne's tamper protection feature, allowing threat actors to disable endpoint detection and response (EDR) agents to install the Babuk ransomware. Like many other software installers, when installing a different version of the agent, the SentinelOne installer terminates any associated Windows processes just before existing files are overwritten with the new version. Threat actors discovered they could exploit this small window of opportunity by running a legitimate SentinelOne installer and then forcefully terminating the install process after it shuts down the running agent's services, leaving devices unprotected.

9. Big Game Ransomware: the myths experts tell board members

   There are many, many examples of organisations paying ransomwares and then taking months to fully restore operations. In fact, in my experience it’s the norm. This is a mistake. Companies do it so they can cover up the breach, which just makes years of problems with regulators.

   It’s time to get serious about building resilient IT systems, locking all the doors, having robust and working containment steps, and to halt the ransomware economic cycle driving these attacks.

10. Making AI models more trustworthy for high-stakes settings

    A new method, called Test-Time Augmentation (TTA), helps convey uncertainty more precisely, which could give researchers and medical clinicians better information to make decisions. Predictions are re-calculated many times, with different ports of the input data excluded. This produces a more accurate estimate of the accuracy of predictions.

11. Claude’s AI research mode now runs for up to 45 minutes before delivering reports

    The upgraded mode enables Claude to conduct "deeper" investigations across "hundreds of internal and external sources," Anthropic says. When users toggle the Research button, Claude breaks down complex requests into smaller components, examines each one, and compiles a report with citations linking to original sources.

    While most reports complete within 5 to 15 minutes, Anthropic says, the new research system can now take up to 45 minutes for particularly complex investigations—tasks that would typically require hours of manual research effort.

    But the results still contain false sources, inaccurate quotations, and other errors.

12. Magento supply chain attack compromises hundreds of e-stores

    A supply chain attack involving 21 backdoored Magento extensions has compromised between 500 and 1,000 e-commerce stores, including one belonging to a $40 billion multinational.

    Sansec researchers who discovered the attack report that some extensions were backdoored as far back as 2019, but the malicious code was only activated in April 2025. The code allows attackers to upload a PHP backdoor, gaining remote code execution.

13. Age Verification in the European Union: The Commission’s Age Verification App

    The European Commission has commissioned an age verification app, to be used for restricted online services. After downloading the app, a user would request proof of their age. For this crucial step, the Commission foresees users relying on a variety of age verification methods, including national eID schemes, physical ID cards with biometrics, etc.

    Once the user would access a website restricting content for certain age cohorts, the platform would request proof of the user’s age through the app. The app would optionally use Zero Knowledge Proofs, a modern advanced cryptographic system that can offer a “yes-or-no” claim (like above or below 18) to a verifier without exposing all the user's information.

    Many jurisdictions in the USA and elsewhere would like to age-gate social media, pornography, and other online resources, but the technical ability to do that has been lacking. This EU system may serve as a model for others, if it works out well. The EFF raises many thorny privacy concerns about age verification, and it will be important to see what the real consequences of this system are.

14. Feeling dumb? Let Google’s latest AI invention simplify that wordy writing for you

    The Google app for iOS has a new "Simplify" feature, which is meant to dumb down complicated writing without losing its meaning. It uses two instances of the Gemini LLM, to create the summary, and then judge its accuracy and readability--a process much like the older Generative Adversarial Network (GAN) used to create images.

15. CL1: The world’s first code deployable biological computer.

    High-performance closed-loop system where real neurons interact with software in real time. A robust environment keeps neurons alive for up to 6 months.

    Biological Intelligence: Traditional AI demands vast datasets, but real neurons learn intuitively with minimal training and a fraction of the energy.

    Science fiction has said "They're keeping Hitler's brain alive in a jar" for decades, finally it's coming true!

16. Why MFA is getting easier to bypass and what to do about it

    An adversary in the middle can defeat most MFA techniques, so we all need to move to WebAuthn, the standard that makes passkeys work. Services that use WebAuthn are highly resistant to adversary-in-the-middle attacks, because WebAuthn credentials are cryptographically bound to the URL they authenticate and to the authorizing device, so the adversary in the middle cannot use them.