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 UsOn Sunday evening at the Difference Makers Awards, we were reminded of why we do the work we do. We honored people and companies who are committed to making the world a better place through compassion, community, and hard work. Dan Kaminsky was remembered and recognized with the DMA Lifetime Achievement award based on his embodiment of these principles and his tremendous contributions to the community. Dan is most widely known for his 2008 discovery of a major vulnerability in the Domain Name System, a critical component underpinning the entire internet, and then for his behind-the-scenes rallying of major DNS providers to work together to create a solution to fix DNS before the vulnerability was announced to the world. He was a cybersecurity researcher and frequent speaker at cybersecurity and hacker conferences, sharing his unbounded interests in all kinds of technology, focusing on making complex topics understandable, and ensuring the security of internet technologies to support human dignity. Dan’s research and presentations included such wide-ranging topics as networking, cryptography, linguistics, pattern analysis, browser attacks, leveraging technology to foster accessibility, and much more. -Ed Skoudis
If you missed it, watch the stream: https://www.youtube.com/live/2Lwch5NEkWs?si=Yj4oJg_CqMWYFmla
Apple released advisories on December 12, 2025, including fixes for two flaws reportedly "exploited in an extremely sophisticated attack against specific targeted individuals," both of which have been added to the US Cybersecurity and Infrastructure Security Agency's (CISA's) Known Exploited Vulnerabilities (KEV) catalog. Both flaws affect WebKit in Apple products before iOS and iPadOS 18.7.3 and 26.2, and version 26.2 of macOS Tahoe, Safari for macOS, tvOS, watchOS, and visionOS. CVE-2025-43529 allows an attacker to carry out memory corruption and achieve arbitrary code execution using maliciously crafted web content, due to a use-after-free vulnerability. CISA notes that CVE-2025-43529 "could impact HTML parsers that use WebKit, including but not limited to Apple Safari and non-Apple products which rely on WebKit for HTML processing." CVE-2025-14174 allows a remote attacker to perform out of bounds memory access using a crafted HTML page due to a flaw in the Almost Native Graphics Layer Engine (ANGLE) graphics library, which is used in both WebKit and Google Chrome. Google also released a patch for this flaw on December 10, tracking the flaw as 466192044 before the CVE was issued, and crediting Apple Security Engineering and Architecture (SEAR) and Google Threat Analysis Group with discovery. Microsoft Edge and Vivaldi browsers have also received fixes for CVE-2025-14174.

In short, Apple released updates for all their OSes and Safari to address the WebKit flaws. If you’re keeping score, this is the 9th zero day for Apple and the 8th for Chrome in 2025. The Apple updates also include bug fixes that your version 26 users will appreciate; use them to sweeten the pot on getting the updates deployed now.
Apple
Help Net Security
The Register
TechCrunch
SecurityWeek
The Hacker News
Three more vulnerabilities have been detected in React Server components. The first vulnerability, CVE-2025-55182, was disclosed with patches available on December 3, and was added to the Cybersecurity and Infrastructure Security Agency's (CISA's) Known Exploited Vulnerabilities (KEV) catalog on December 5. While this flaw initially had a mitigation deadline of December 26, CISA moved the deadline up two weeks to December 12 due to widespread reports of active exploitation. The additional vulnerabilities are CVE-2025-55184, a high-severity deserialization of untrusted data issue; CVE-2025-67779, a high-severity issue detected when it was determined that the fix for CVE-2025-55184 was incomplete; and CVE-2025-55183, a medium-severity vulnerability that could lead to information leaks.

Time to paint the bridge again. While it’s easy to armchair quarterback, the reality is they didn’t get a full fix in the effort to respond quickly to the flaw. Even so, the further issue was rapidly addressed. Roll out the update, then see how your developers would handle a scenario like this. Make sure they have the top cover to execute successfully.
React
The Register
The Register
Heise
The Hacker News
The Hacker News
Palo Alto Networks
Wiz
NIST
NIST
NIST
NIST
The US Cybersecurity and Infrastructure Security Agency (CISA) has added a high-severity improper restriction of XML external entity reference vulnerability in GeoServer (CVE-2025-58360) to the Known Exploited Vulnerabilities (KEV) catalog. In November, GeoServer maintainers noted that "The application accepts XML input through a specific endpoint /geoserver/wms operation GetMap. However, this input is not sufficiently sanitized or restricted, allowing an attacker to define external entities within the XML request." The flaw affects all versions of GeoServer up through 2.25.5 as well as versions 2.26.0 through 2.26.1. The vulnerability has been fixed in versions 2.25.6, 2.26.2, 2.27.0, 2.28.0, and 2.28.1. The Canadian Centre for Cyber Security has updated its advisory on the GeoServer vulnerability to note that an exploit for the flaw exists in the wild. The vulnerability has a mitigation due date of January 1, 2026 for US Federal Civilian Executive Branch agencies. Shodan has observed more than 14,000 exposed GeoServer instances. CISA has also added a high-severity unrestricted file upload vulnerability in Sierra Wireless AirLink ALEOS routers (CVE-2018-4063) to the KEV catalog. Cisco Talos reported the vulnerability to Sierra, which is based in Canada, in December 2018; the vulnerability was publicly disclosed in April 2019. Because the product has reached end-of-support, FCEBs are asked to discontinue its use or update to a supported version by January 2, 2026. Other recently added vulnerabilities include a hard-coded credentials issue in Gladinet CentreStack and Triofox (CVE-2025-14611), and two flaws covered in this issue of NewsBites: an out-of-bounds memory access flaw in ANGLE in Google Chrome on Mac prior to 143.0.7499.110 (CVE-2025-14174), and a use-after-free vulnerability in WebKit affecting multiple Apple products (CVE-2025-43529).

The Sierra flaw is six years old, and you should have applied the update long ago. Even so, if you’re running their LS300, GX400, GX440, ES440, GX450, ES450, MP70, MP70E, RV50, RV50X, LX40, or LX60 devices, you’re on the patched, supported firmware. Also make you’re using strong authentication, limiting access to the management interface and that management is performed over SSL.

The value of intel is proportional to its use.
SC Media
SecurityWeek
NIST
GitHub
Government of Canada
The Hacker News
NIST
MITRE has published its annual 25 Most Dangerous Software Vulnerabilities report. The list is compiled based on analysis of vulnerabilities present in more than 39,000 CVEs. Cross-site scripting (XSS) occupies the top spot on MITRE's 2025 list, followed by SQL injection and cross-site request forgery (CSRF), missing authorization, and out-of-bounds write. Other vulnerabilities appearing on the list include code injection, OS command injection, path traversal, out-of-bounds read, and use-after-free. MITRE has also published a methodology page to clarify how the list was compiled.

When I see a list like this, I try to find themes to simplify the message to others. Input validation, sanitization, and misplaced trust are recurring issues across these CWEs. Be dogged in making sure all external input is checked, no matter the source. Another thread is authorization and access control. Watch for end arounds and weak authentication. Strong MFA is essentially turnkey these days, so your developers can focus on implementing good ACLs. Make sure you have process to validate access granted, removing it when no longer needed.
Good analysis by MITRE. What it really speaks to is a lack of discipline in secure software development within the Industry. The Center for Internet Security (CIS) and SAFECode recently released a guide to assessing software security practices. The guide is based off of NIST’s Secret Software Development Framework (SSDF) and is a call to action for organizations to embed Secure by Design principles in corporate strategy and culture, regularly assessing software assurance maturity.

39,000 CVEs, almost none of which will ever be completely patched, perhaps not even every instance of these 25.
MITRE
MITRE
MITRE
BleepingComputer
SecurityWeek
Microsoft has announced this it is deprecating the Rivest Cipher 4 (RC4) encryption cipher that Windows has supported since 2000 with the rollout of Active Directory. Despite the fact that the cipher's source code was leaked in 1994, RC4 remained in regular use for another two decades. While Microsoft Active Directory has been updated to support AES, "Windows servers have continued to respond to RC4-based authentication requests and return an RC4-based response" by default, making it easier for attackers to steal account credentials. By the middle of next year, Microsoft "will be updating domain controller defaults for the Kerberos Key Distribution Center (KDC) on Windows Server 2008 and later to only allow AES-SHA1 encryption. RC4 will be disabled by default and only used if a domain administrator explicitly configures an account or the KDC to use it."

While your first reaction may be to just disable RC4, you’ll want to see if it’s in use first. Microsoft has added fields to Kerberos events 4768 and 4769 to allow you to detect RC4, and have published PowerShell scripts that read this information, so you know which accounts are still RC4 only and which don’t support AES. The goal is to make sure accounts and devices are enabled for AES, then disable RC4.

While quantum computing may reduce the time to break an RSA key to hours or days as early as 2030, RSA will likely last forever. We create tens of thousands of RSA keys per hour. Cryptography fails slowly, not suddenly. The advantage is always to the cryptographer, never to the cryptanalyst. The cost to the cryptographer rises linearly while that to the cryptanalyst rises exponentially.
Jaguar Land Rover (JLR) has confirmed that the perpetrators of the August 2025 cyberattack stole personal data from the company's payroll system, affecting both current and former employees. JLR told The Telegraph, "From the ongoing forensic investigation, JLR believes that certain data related to current and former JLR employees, and contractors, was affected by the cyber incident. We remain in dialogue with the relevant regulators and we are in the process of contacting current and former employees and contractors as necessary." The breach has resulted in significant losses for the company, and is predicted to have repercussions on the UK economy overall.

JLR isn’t out of the woods yet. They prioritized getting systems and production back online, which is likely how their DR plan is structured. Make sure you know what the priorities are in your plan, to include notifications of those affected by loss of sensitive information. As a consumer, recognize that letting you know your information is lost may not be priority one for a company; take steps to protect yourself accordingly.
The hits they keep a-coming for JLR. This cyber incident will most certainly be used as a case study for MBA students for years to come. Hopefully it will be a teachable moment for future company leaders.

This breach put the health of the enterprise, the life of some supplier enterprises, and the productivity of the nation at risk. Compromise of PII is way down the list of consequences.
Data security breaches at Prosper and 700Credit have compromised personal information belonging to roughly 19 million people. Prosper Marketplace, a peer-to-peer lending firm, detected a breach on September 1, 2025, and notified regulators; intruders had access to company systems between June and August 2025. The Prosper breach affected more than 13 million people, and the compromised data include general identification data as well as financial account, payment card, and tax data. 700Credit, a company that provides credit reports, ID verification, and related services to car dealerships, will begin notifying approximately 5.8 million people that their personal information, including basic identification information including Social Security numbers, was compromised. The 700Credit breach occurred through an integration where the intruder discovered an API that allowed access to customer information. The partner company did not inform 700Credit of the incident. Instead, 700Credit detected unusual activity on its systems in late October and launched an investigation.

Takeaways from the 700Credit incident include the importance of understanding incident notification processes for third parties as well as verification of security practices, particularly API security, which is harder with a third party. Make sure you understand what your obligations are for a third-party breach; your customers likely don’t care where the fault lies, and will look to you to be made whole, so you need this squared away with finance, legal, and management before you need it. Then double check how frequently you’re verifying the security of your third-party providers, to include verification of the relationships.
First off, 19M: that’s a big number for two cyber incidents. APIs by definition are designed for integration. With that integration comes complexity, which greatly increases one’s attack surface. The 700Credit cyber incident is a good reminder about supply chain attacks and the need for incident reporting to be part of the Service Level Agreement (SLA).
Police in Seoul, South Korea are investigating a breach at online mega-retailer Coupang. The company suffered a breach in late June 2025 that compromised data belonging to 33.7 million customers. The incident was not detected until November 18 and was disclosed on December 1. The breach has been attributed to a former employee who had worked on an authentication management system and maintained access to an internal system following departure from the company. Police raided Coupang offices as part of their investigation, gathering digital forensic evidence, including documents, logs, system records, IP addresses, access credentials, and access history. While the breach has been attributed to the unnamed individual, Coupang could be held liable if the company is found to have been negligent.

Here is a good example of why you want timely, comprehensive, account management. This may be a good scenario to bring up if your point isn’t being appreciated. Validate that your team knows how to act in the event of a legal or regulatory investigation of a problem; while these aren’t fun, it’s easy to make things worse.
Yet another reminder for companies to have a well-rehearsed process in place for the removal of employees from employment. The documented process should engage direct supervisors, HR (Talent Management), IT, and Legal. Most importantly, immediately upon notification of employee removal, access to the building and IT systems should be restricted. As far as whether Coupang is liable, the court will decide. That said, failing to remove access to all company systems upon termination is not demonstrating reasonable cybersecurity.
France's Minister of the Interior, Laurent Nuñez, stated to French media outlet RTL that a cyberattack against the country's Ministry of the Interior took place overnight from December 11-12, 2025. The attacker accessed email servers and certain files, but Nuñez contends that there is no evidence of serious compromise; whether information was stolen has not yet been determined. The ministry is responding by enacting its standard protection procedures, and is working to strengthen security and tighten system access. The nature, scope, and origin of the attack remain under investigation. BleepingComputer notes that the Ministry of the Interior "supervises police forces and oversees internal security and immigration services."

They are in an interesting spot: a successful attack occurred and they are taking measures to tighten security which may impact users, so disclosing the event is warranted. The current notification is lacking in impact as well as in instructions or contacts for those impacted. Make sure your notifications are clear and include this information.
RTL
BleepingComputer
Reuters
SANS Internet Storm Center StormCast Tuesday, December 16, 2025
Current React2Shell Example; SAML woes; MSMQ issues after patch
https://isc.sans.edu/podcastdetail/9740
More React2Shell Exploits CVE-2025-55182
Our honeypots continue to detect numerous React2Shell variants. Some using slightly modified exploits
https://isc.sans.edu/diary/More+React2Shell+Exploits+CVE202555182/32572
The Fragile Lock: Novel Bypasses For SAML Authentication
SAML is a tricky protocol to implement correctly, in particular if different XML parsers are used that may not always agree on how to parse a specific message
https://portswigger.net/research/the-fragile-lock
December Updates Causes Issues with Microsoft Message Queuing
SANS Internet Storm Center StormCast Monday, December 15, 2025
DLL Entry Points; ClickFix and Finger; Apple Patches
https://isc.sans.edu/podcastdetail/9738
Abusing DLLs EntryPoint for the Fun
DLLs will not just execute code when some of their functions are called, but also as they are loaded.
https://isc.sans.edu/diary/Abusing+DLLs+EntryPoint+for+the+Fun/32562
Apple Patches Everything: December 2025 Edition
Apple released patches for all of its operating systems, fixing two already exploited vulnerabilities.
https://isc.sans.edu/diary/Apple+Patches+Everything+December+2025+Edition/32564/
ClickFix Attacks Still Using the Finger
Two examples of ClickFix attacks abusing the finger protocol to load additional malware
https://isc.sans.edu/diary/ClickFix+Attacks+Still+Using+the+Finger/32566
Denial of Service and Source Code Exposure in React Server Components
After last week's critical patch, three more, but less critical, vulnerabilities were identified in React Server Components.
Catch up on recent editions of NewsBites or browse our full archive of expert-curated cybersecurity news.
Browse ArchiveDiscover how integrating AEM with ServiceNow delivers the missing link between visibility and control, connecting live endpoint intelligence directly into your existing IT workflows. And hear how Tanium and their partners like AHEAD make adoption simple, guiding design, orchestration, and long-term operational success.
Webinar | Securing Remote Access in OT: A Critical Control for Modern Risk | Wednesday, January 14th @ 1:00 PM ET. With SANS Certified Instructor Michael Hoffman.
Take the SANS 2026 CTI Survey. Share your knowledge with the community and be entered into a drawing to be one of four $100 Amazon gift card winners.
Free Virtual Event | SANS 2026 Kubernetes & CNAPP Forum | January 15, 2026 @ 10:00 am ET. | 4 CPE credits.