SEC595: Applied Data Science and AI/Machine Learning for Cybersecurity Professionals


Experience SANS training through course previews.
Learn MoreLet us help.
Contact usBecome a member for instant access to our free resources.
Sign UpWe're here to help.
Contact UsAs of March 15, 2026, the maximum certificate lifespan for TLS certificates has dropped to 200 days. The change is one step in a process established by the CA/Browser Forum: On March 15, 2027, maximum validity period of subscriber certificates will drop to 100 days, and on March 5, 2029, that window will be shortened to 47 days. Certificate authorities began making the change shortly before the deadline. DigiCert moved to issuing certificates with a maximum 199-day validity window on February 24, 2026, and SSL began issuing certificates with 200-day expiration dates on March 11, 2026. As SSL explains, the rationale for the shortened validity window lies in enhanced security, improved cryptographic agility, and stronger validation practices.

The lifetime of TLS certificates will continue to decrease. This is your chance to get your automation working, unless you intend to retire by the end of the year, in which case you may as well have the next person fix it. But there is also a chance of leaving behind a great puzzle in the form of an undocumented script.

This is the continuation of an ever-shrinking certificate lifetime. Your mission — and whether or not you accept it, these shortened lifetimes are coming — is to automate your certificate updates. You want to be in a position where you don't have to care how rapidly certificates need to be updated. With the increasing number of places we have certificates, and with shortening timelines, hand feeding and maintenance of them is becoming untenable/unsustainable. Start with discovery: You need to know where all your certificates are, as well as what encryption is in use, and you need to be prepping for PQC as well. Start with low-hanging fruit and save the more difficult areas, hopefully less than 20%, for after these are all squared away and you've got more experience to draw from.

The CA/Browser forum should track and publish metrics showing impact of the "stronger validation practices" that should occur at issuing CAs. Doing shoddy validation more frequently does NOT change anything. Look out for a wave of AI ("Ample Inaccuracies")-based registration approaches…
Misused or mismanaged digital certificates are a concern; moving to a 47-day lifecycle greatly limits that risk. In order to balance the increased operational cost, automating the process will be essential. All in all, it is the right thing to do.

Perhaps my colleagues can help me understand how, beyond a certain point measured in months rather days or weeks, increasing the frequency of certifying improves security. It will increase revenue and may increase profits, depending upon how much the certifying authority spends vetting applications, but it is hard to see how it improves security.
Help Net Security
CABForum
DigiCert
SSL
StepSecurity is warning of an ongoing campaign compromising hundreds of Python repositories on GitHub with GlassWorm malware, which notably is now being injected into developers' code without leaving evidence in the activity feed. GlassWorm — which takes its name from the invisible Unicode characters that hide the malicious code — was first observed in October 2025 in extensions from the OpenVSX marketplace; it uses Solana Blockchain for command-and-control (C2) and installs an infostealer, then uses developers' stolen tokens to take over accounts and propagate itself into additional repositories. "Anyone who runs pip install from a compromised repo or clones and executes the code will trigger the malware." The initial infection in this March 2026 campaign is through malicious VSCode and Cursor extensions. StepSecurity calls the new injection technique "ForceMemo," because the attacker finds the latest legitimate commit of a branch, rebases it to include the GlassWorm code, and then force-pushes to the default branch, with Solana transaction memos as the C2 channel. This preserves the original commit message and author date, and sets the committer email to "null." The malware loader is also added as an extension dependency, rather than being included in the package. "The targeted repos include Django web applications, machine learning research code, Streamlit dashboards, Flask APIs, and Python packages installable via pip install from GitHub URLs." Aikido and Socket have also published notices, noting that the same technique is being used in npm and VS Code marketplace, and that most, but not all infected OpenVSX extensions have been removed. Users should follow StepSecurity's instructions to check for indicators of compromise and review git commit history.

The attack begins with compromised accounts from developers who were infected with the GlassWorm malware, grabbing their ~/.git-credentials and GITHUB_TOKEN environment variables. Next the malware is added using a technique which causes the new commit to not appear as new, as the commit comment and author date remain unchanged, but the committer date is still updated. Grab the IoCs from the StepSecurity blog, check for the marker variable as well as ~/init.jason which provides persistence, and review your git commit history of repos you've cloned. Keep an eye out for code comments in Russian, Solana C2 address, and funding wallet.

It used to be sufficient to say, "get software from trusted sources in tamper-evident packaging." Today, not so much.
StepSecurity
Aikido
Socket
Ars Technica
SecurityWeek
The Hacker News
Dark Reading
Google has updated Chrome to fix two high-severity vulnerabilities that are being actively exploited. CVE-2026-3909 is an out-of-bounds write in Skia in Google Chrome prior to 146.0.7680.75 that allowed a remote attacker to perform out of bounds memory access via a crafted HTML page. CVE-2026-3910 is an inappropriate implementation in V8 in Google Chrome prior to 146.0.7680.75 that allowed a remote attacker to execute arbitrary code inside a sandbox via a crafted HTML page. Users are urged to ensure that they are running the most current versions of Chrome, which are 146.0.7680.75/76 for Windows/Mac and 146.0.7680.75 for Linux at the time of writing. Both vulnerabilities were discovered in-house, and both have been added to the US Cybersecurity and Infrastructure Security Agency's Known Exploited Vulnerabilities (CISA KEV) catalog with mitigation deadlines of March 27, 2026.

I was wondering about that update button I spotted today. Hopefully all your browsers are automatically checking for updates and installing them on next launch. That means all you need to do is relaunch regularly. And while you're at it, don't forget to check your non-Chromium based browsers, such as Firefox; even with less frequent updates, your discipline should be equivalent.

Interestingly, Google later removed one of the zero-days from the description and noted that it will be fixed in a future release.
We’re at the point where knowing that Chrome has patched a zero-day is less important. I mean, you should already have a process in place to close and restart your browser on a daily basis. It’s just a good user behavior to instill, that has an instant security benefit.

Browser vulnerabilities have many instances and many applications per instance. Prefer purpose built clients for sensitive applications.
Heise
The Register
The Hacker News
BleepingComputer
GoogleBlog
CVE
CVE
Hewlett Packard Enterprise (HPE) has published a security advisory warning of five vulnerabilities, one rated critical, in HPE Aruba Networking AOS-CX before versions 10.10.1180, 10.13.1161, 10.16.1030, and 10.17.1001. CVE-2026-23813, CVSS score 9.8, allows an unauthenticated remote attacker to circumvent authentication controls for AOS-CX network switches, including resetting the admin password, due to a flaw in the web-based management interface. Three of the flaws are high-severity: a command parameters flaw allowing malicious command injection (CVE-2026-23814), a flaw in a custom binary in the CLI allowing command injection (CVE-2026-23815), and a flaw in the CLI allowing command injection at the operating system level (CVE-2026-23816). The fifth flaw is rated medium, and allows an attacker to redirect users to an arbitrary URL due to a vulnerability in the web-based management interface. Beyond applying the patches, HPE recommends the same initial mitigation for all five vulnerabilities: ensure that "management interfaces be restricted to a dedicated layer 2 segment/VLAN and/or controlled by firewall policies at layer 3 and above, along with accounting controls for tracking and logging user activities and resource usage." For the critical flaw, HPE has additional recommendations: "Disable HTTP(S) interfaces on Switched Virtual Interfaces (SVIs) and routed ports wherever management access is not required. Enforce Control Plane Access Control Lists (ACLs) to protect any REST/HTTP-enabled management interfaces, ensuring only trusted clients are allowed to connect to the HTTPS/REST endpoints. Enable comprehensive accounting, logging, and monitoring of all management interface activities to detect and respond to unauthorized access attempts promptly."

This affects the web-admin interface of your Aruba CX switches and can be remotely exploited without authentication. You need to do a couple of things: First, make sure that you're on the supported updated release. Older, unsupported, versions of AOS-CX have the flaw but will not be getting updates. Second, make sure that you're restricting access to the web-admin interfaces to only authorized devices, preferably not over the WAN. Two more things: Make sure that you've enabled comprehensive logging, and disable the HTTPS interfaces on switched virtual interfaces and routed ports.

Looking at the bug descriptions, I was surprised that these are 2026 CVEs, and not 1999 CVEs. But then I noticed that this is an expensive enterprise product, so the 2026 CVE numbers make sense.
The UK government's business registry, Company House, says that a vulnerability in its WebFiling service could have allowed "a logged-in user of [the] WebFiling service [to] potentially access and change some elements of another company’s details without their consent after performing a specific set of actions." Companies House, which oversees the incorporation and dissolution of companies, temporarily shuttered the WebFiling service on Friday, March 13, and the service remained closed over the weekend. The issue has now been resolved, and the service is online as of Monday, March 16. Company House became aware of the vulnerability through Dan Neidle, a tax professional, who learned of the issue from Ghost Mail's director of operations John Hewitt. The incident has been reported to the Information Commissioner’s Office (ICO) and the National Cyber Security Centre (NCSC). The issue appears to have been introduced in an October 2025 WebFiling update.

On a system affected by the flaw, once logged in, you looked up another company by entering its company number, hit the access button on that screen, and at the authentication token prompt, hit the back button four times and instead of returning to your dashboard, you were logged into the dashboard of the company you're trying to access. The user who discovered the flaw didn't receive any acknowledgement, so the founder of Tax Policy Associates reported the flaw to the UK corporate register. Two lessons here: First, don't disregard reported flaws. Second, robust testing of code changes and access controls is still necessary. It is likely that access tokens were set prior to authentication being completed — a reminder that order of operations really matters.
The Register
Gov Infosecurity
Infosecurity Magazine
BleepingComputer
UK Gov
Vancouver, British Columbia-based *TELUS Digital* has acknowledged that it experienced a cybersecurity incident after threat actors claimed to have stolen nearly one petabyte of data from the business process outsourcing company. TELUS Digital is investigating the incident but has not provided any specifics about when the incident occurred, nor what kind of data were compromised. Canadian retailer *Loblaw* has disclosed a data security breach that compromised basic customer data, including names, phone numbers, and email addresses. Loblaw, which is based in Brampton, Ontario, has notified customers of the incident, and acknowledged that as a precaution, all customers will be logged out of their accounts.

The ShinyHunters gang is taking credit for the TELUS Digital breach, claiming they breached their Google Cloud Platform using credentials stolen during an earlier Salesloft Drift breach. Loblaw is a Canadian retailer with 2,400 locations operating brands including Shoppers Drug Mart, No Frills, Real Canadian Superstore, and President's Choice. While passwords weren't compromised, if you're a Loblaw user it would be a good idea to change your password regardless.
Bleeping Computer
Telus Digital
Loblaw
SecurityWeek
BleepingComputer
An international law enforcement operation coordinated by INTERPOL has arrested 94 individuals for their alleged roles in phishing and ransomware operations. Another 110 individuals remain under investigation. The operation also took down more than 45,000 malicious IP addresses and seized more than 200 electronic devices and servers. Operation Synergia III involved law enforcement agencies from 72 countries and territories as well as private sector partners. The operation ran from mid-July 2025 through the end of January 2026. The criminal activity included websites used in ransomware and phishing; phony sites for casinos, government, banking, and payments services; and romance scams, loan and job scams, payment card fraud, and identity fraud.

This is an example of how international cooperation can really succeed. The operation has been growing since 2023’s Synergia I, with 52 countries participating and seizing 1,300 IP addresses, to Synergia III involving 72 countries and seizing 45,000 IP addresses. They still have many more individuals under investigation; expect to hear more from Operation Synergia.
Interpol
The Register
Help Net Security
Infosecurity Magazine
The Hacker News
In a statement on March 12, 2026, Poland's National Centre for Nuclear Research (NCBJ) said that "an attempted cyberattack on the Institute's IT infrastructure" was thwarted "thanks to the rapid and effective actions of security systems and procedures ... as well as the quick response of our teams." NCBJ says that production, operations, and research were not disrupted and that "the MARIA nuclear reactor is safe." According to Reuters, NCBJ "is Poland’s main government nuclear research institute specializing in nuclear physics, reactor technology, particle physics, and radiation applications. It provides technical and scientific support for the country’s nuclear power program." Poland is building the country's first nuclear power plant. In January, elements of Poland's power grid — distributed energy resource (DER) sites, heat and power (CHP) facilities, wind, and solar dispatch systems — were also targeted by cyberattacks.

Not a lot of detail on how the attack was thwarted but good to see low Time to Detect and Time to Mitigate highlighted. There is no shortage of failure stories — we need to have more case studies on successes to enable others to more easily follow similar paths.

The takeaway is that early detection and response works. The thing is, you need to have not only the security infrastructure in place and operating, but also tested response plans. Purchasing shelfware, lacking alerts to events, or having staff not trained to respond to those alerts, can all result in your undoing, regardless of the threat actor. Don't be that target.
Recently unsealed court documents name Angelo Martino as the previously unnamed co-conspirator of Ryan Clifford Goldberg and Kevin Tyler Martin in extortion of US companies as affiliates of the ALPHV BlackCat ransomware scheme in 2023. At the time of the attacks, Martino and Martin were employed as ransomware threat negotiators at DigitalMint, and Goldberg was an incident response manager for Cygnia Cybersecurity Services. Martin and Goldberg were indicted in October 2025 and pleaded guilty to charges of extortion in December. The trio is variously indicted for ten ransomware attacks, six of which resulted in ransomware payments: one paid to Martin, Goldberg, and Martino directly, and five paid to other unnamed ALPHV BlackCat co-conspirators when Martino exploited his position as ransomware negotiator in exchange for a portion of the ransom, providing "direction and confidential information" about DigitalMint clients to the attackers to maximize the ransom payment. The total paid in ransoms across these cases exceeds US$75.25 million, with additional losses for all victims from the theft and encryption of their data. Martino's plea will be heard on March 19; in the meantime, his properties, vehicles, and financial assets have been seized, and he is forbidden from leaving the Southern District of Florida and from ever working in the cybersecurity industry again. The maximum term of imprisonment for these charges is 20 years.

We finally know who the third musketeer is, and that he was also working for DigitalMint with Kevin Martin. I can't help but feel this calls into question any plans of negotiating for a ransom demand, particularly as the guidance remains not to pay, instead of enlisting agencies like the FBI to obtain a resolution and decryption option.
Document Cloud
The Record
HIPAA Journal
BleepingComputer
SANS Internet Storm Center StormCast Tuesday, March 17, 2026
Proxy URLs; Local Network Address Restrictions; Advanced Phishing
https://isc.sans.edu/podcastdetail/9852
/proxy/ URL scans with IP addresses
https://isc.sans.edu/forums/diary/proxy+URL+scans+with+IP+addresses/32800/
Local Network Address Restrictions
https://learn.microsoft.com/en-us/deployedge/microsoft-edge-relnote-stable-channel
European Security Vendor Targeted by Hackers Fronting as Cisco Domain
https://specopssoft.com/blog/phishing-campaign-cisco/
SANS Internet Storm Center StormCast Monday, March 16, 2026
SmartApeSG and Remcos RAT; React-based Phishing; Google Chrome Patches; AdGuard Vuln
https://isc.sans.edu/podcastdetail/9850
SmartApeSG campaign uses ClickFix page to push Remcos RAT
https://isc.sans.edu/diary/SmartApeSG+campaign+uses+ClickFix+page+to+push+Remcos+RAT/32796
A React-based phishing page with credential exfiltration via EmailJS
https://isc.sans.edu/diary/32794
Google Chrome announced two zero-day fixes, then removed one.
https://chromereleases.googleblog.com/2026/03/stable-channel-update-for-desktop_12.html
AdGuard Vulnerability
https://github.com/AdguardTeam/AdGuardHome/releases/tag/v0.107.73
Catch up on recent editions of NewsBites or browse our full archive of expert-curated cybersecurity news.
AI-Accelerated Threat Landscape: Year of the Evasive Adversary. Join us for an exclusive virtual threat briefing on the CrowdStrike 2026 Global Threat Report, where experts will reveal how threat actors are evading traditional defenses by weaponizing AI, targeting unmanaged edge devices, and exploiting cross-domain blind spots at record speeds.
Live Sessions Today | SANS Leadership Summit Solutions Track 2026 | Tuesday, March 17, 2026, at 10:00 AM ET.
Webinar | AI-Human Collaboration in Modern SOCs | Wednesday, March 18, 2026, at 3:30 PM ET.
Webinar | The Multicloud Blueprint: Architecting Security for the AI Era | Tuesday, March 31, 2026, at 1:00 PM ET. https://www.sans.org/info/235230