Too many organizations are STILL getting breached. Every day across large and small organizations, intrusions and breaches happen.

Every day across organizations both large and small, intrusions and breaches happen. Attackers get inside. If the organizations are fortunate, they detect and get them out before they do any damage. They remediate the situation before the intrusion turns into an official breach. But for many less fortunate, when breaches happen they can last for weeks, months, or years under the radar. Once finally discovered, the investigations can be long and painful, and they often get publicized.

We live in a world where attackers appear to have the upper hand and, on some days, even seem to be winning. It's hard to understand the current state of affairs when there is an endless number of cybersecurity vendors, service providers, and experts touting their abilities to secure organizations of all sizes.

There are many promises. Many promote 99.9% accuracy and their ability to stop all breaches. Vendors talk about their solutions having artificial intelligence (AI) and machine learning (ML) to identify unknown threats, but not too many people can really explain exactly how AI and ML work in cybersecurity. There's a lot of hype.

There is not a single vendor on the planet right now that can provide a one-stop shop of world-class technology to prevent and stop breaches. One doesn't exist. Organizations need to be able to choose best-in-class technologies that work well and integrate together no matter what company built them.

**Breaches Keep Happening**

According to the Identity Theft Resource Center, the landscape has not improved much over the last 15 years. With all of the protection and intelligence available contrasted against successful intrusions and breaches, something is not adding up.

The industry as a whole has not achieved the objective of preventing, or even mitigating, breaches.

We must keep in mind that while intrusions and breaches are a reality, they don't need to be devastating. One of the main reasons they often are so harmful: blind spots.

Despite security controls focused on specific areas of environments such as identity and access management (IAM), endpoint protection platform (EPP), endpoint detection and response (EDR), next-generation firewall (NGFW), data loss prevention (DLP), network detection and response (NDR), and so on, blind spots are still everywhere. All these different security controls are great for looking at the area they're assigned, but if they aren't all talking to each other, organizations are flying blind.

**Attackers Love Blind Spots and Credentials**

While security teams are chasing false alerts, external attackers are finding legitimate credentials already exposed, and exploiting vulnerabilities that enable them to find credentials from within the environment. Or they're using a large amount of money to entice a legitimate user to share their credentials voluntarily. Once the credentials are in hand, a bad actor can take their time to scour the environment, map sensitive data locations, and quietly create "backdoors" for future use.

If the attacker is more of the "smash and grab" type, they can carry out a flash attack, deploy malware, ransomware, or any number of damaging attacks and watch the chaos ensue.

For those rare trusted employees who goes rogue, their path to carrying out a devastating attack is much shorter. Already with an established presence, legitimate access, and user IDs/passwords inside the environment, the opportunity to prevent them in carrying out nefarious activities is often nonexistent. The only hope for organizations is the domain of detection and response.

**Know Normal, Prevent, and Detect **

Security teams need to know what is normal behavior in their organization to quickly identify anything abnormal like the situations mentioned above. Right now, there is still way too much focus in cybersecurity on prevention, and not enough on detection and response. No matter how many prevention tools are in place, attackers are still getting in and insiders are still getting out. Too many security operations teams are still flying blind.

Currently, organizations will continue to experience intrusions and breaches, but what the pain and lasting consequences aren't inevitable. By incorporating the ability to determine what normal activity is for users and entities, organizations stand a better chance of detecting the abnormal and uncovering external and insider threats (whether malicious or accidental), turn the tables on the attackers, and mitigate damage. And that's true even as "normal" constantly changes.

Organizations will win when they know normal and identify what's abnormal — the breach.

**About the Author**

    

Gorka Sadowski is Chief Strategy Officer at Exabeam. In his role, Sadowski assists the executive team and functional leaders across the company. Sadowski has more than 30 years of security experience. Most recently, Sadowski was senior director and security and risk management analyst at Gartner. Prior to Gartner, Sadowski led business development at Splunk and built the Splunk security ecosystem. Prior to Splunk, Sadowski established presence for LogLogic in southern Europe, ran security activities for Unisys in France, and launched the first partner-led intrusion detection and prevention system in the industry.

Flying Blind in Security Operations

Trying to remediate everything was never a winning strategy. RBVM is an approach that gets organizations better results with less effort.

For the past five years, the National Vulnerability Database (NVD) has broken its own record of reported vulnerabilities and is on pace to do the same in 2022. With a threat landscape growing that quickly, it's no surprise to see security teams can't keep pace.

According to a report from Cobalt, 79% of security teams struggle to consistently monitor for vulnerabilities amid today's labor shortages, and 54% of security respondents are considering quitting their jobs.

Trying to remediate everything was never a winning strategy. Risk-based vulnerability management (RBVM) is an approach that gets organizations better results with less effort than trying to keep up with the daily average of more than 60 new vulnerabilities so far in 2022.

Five years of data backs this up. The top four metrics used to measure success with RBVM have consistently gotten better, even as a record number of common vulnerabilities and exposures (CVEs) has added more noise to the fray each year.

Let's take a deeper look at those metrics, how they've improved, and why they've improved.

**RBVM 101**

In its early days, RBVM was taboo because everyone saw vulnerability management as an all-or-nothing proposition. Precious few reached "all" status, and getting there meant a lot of time wasted on remediating vulnerabilities that posed no threat.

"Maturity models" attempt to collect the best cybersecurity practices determined by experts and tweak them over time, but there's an even better way: data. A series of research reports from the Cyentia Institute has provided some answers on what organizations should be prioritizing and informed a better approach to RBVM.

Research produced in conjunction with the Cyentia Institute has shown that 23% of published vulnerabilities have associated exploit code and 2% have observed exploits in the wild. That dramatically reduces the amount of effort required … if you're remediating the right way.

We can judge how organizations are doing by looking at remediation capacity, velocity, coverage, and efficiency. The Cyentia Institute analyzed data from Kenna's platform across the past five years and visualized it in the charts below.

Capacity, the proportion of open vulnerabilities being closed on average every month, has generally increased over the past five years for leading and average organizations. While most organizations are remediating more of their open vulnerabilities, the bottom quarter of organizations has remained steady with no meaningful shifts.  

    

Source: Kenna Security

Remediation velocity saw quick, meaningful progress and has since been improving slowly. Where we are now is a _vast_ advancement over early 2016, when the half-life was more than 125 days. Over the last four years, the half-life — or time it takes to remediate half the newly discovered vulnerabilities — has dropped by more than a third from 32 days to 21 days.  

    

Source: Kenna Security

Efficiency and coverage are related, which is why they're shown together, and without a significant change in prioritization strategy, improving one of those metrics will often lead to a decrease in the other. For example, one way to increase coverage is to simply remediate more, but that makes you less efficient.

Around 2018, both coverage and efficiency were around 35%, but an increase in coverage (more exploited vulns are remediated correctly) causes a decrease in efficiency (fewer of the remediated vulnerabilities are being exploited). This is likely caused by organizations increasing their capacity to remediate more vulnerabilities.  

    

Source: Kenna Security

Keep in mind that this is all happening as we've seen the number of reported vulnerabilities increase each year since 2017. Organizations that have adopted RBVM are filtering out the noise by focusing on risk. They're also remediating more vulnerabilities faster, which is why coverage and efficiency continue to improve.

**Explaining the Success**

While a more intelligent approach to vulnerability management makes the biggest difference, vendors have gotten smarter, too. Research tells us that Microsoft has more vulnerable assets observed and more exploitation activity than its peers. When the software giant issues patches quickly, it has a heavy impact on remediation velocity (organizations address half their vulnerabilities affecting Microsoft products in roughly 22 days, more than 40 times quicker than SAP or Linux at 900 days).

Between a smarter approach and quick patches from software vendors, vulnerability management is in a far better position than it was five years ago. The proof is in the numbers that organizations have better success when they approach security through a risk lens.

The labor market has further complicated the already complex issue of security, but now organizations know they don't have to fix everything. In fact, it would be silly to and you'll end up driving your valuable security practitioners away. By adopting RBVM, organizations can start to quell the unrelenting pandemonium today's security landscape has become.

How Risk-Based Vulnerability Management Has Made Security Easier

Ducktail targets marketing and HR professionals through LinkedIn to hijack Facebook accounts and run malvertising schemes.

A spear-phishing campaign dubbed "Ducktail" has been discovered targeting marketing and HR professionals through LinkedIn, with the aim of taking over Facebook Business accounts and abusing the Ads function to run malvertising schemes.  

The campaign delivers a tailored malware, which identifies individuals likely to have admin privileges, scans the victim' machine, searches for popular browsers, and extracts all the stored cookies, including any Facebook session cookies, from the browsers it finds.  

The malware component can take advantage of authenticated Facebook sessions to steal information from the victim's Facebook account, which it then uses to hijack the victim's Facebook Business and Facebook Ads accounts.

The campaign appears to be financially motivated, according to security specialist WithSecure's new report on the Ducktail campaign.

"One of the unique features of the malware is its ability to hijack Facebook Business accounts associated with the victim's Facebook account," WithSecure's report explains. "It attempts to grant the threat actor's emails access to the business with the highest-privilege roles."

Mohammad Kazem Hassan Nejad, researcher for WithSecure Intelligence, explains that the attackers carefully select their targets, making sure they're likely to be on Facebook Ads or Business first. if the threat actor blindly distributes the malware through other forms of attack, such as malicious spam campaigns, this could ring more bells that would alarm companies, cybersecurity vendors, and Meta about Ducktail's activity much sooner, he notes.

"By scouting for companies that operate on Facebook's Ads and Business platform beforehand and targeting individuals that most likely have access to a Facebook Business account, we believe the threat actor tries to increase their chance of success whilst making the least amount of noise," he says.

**Connections to SilentFade**

Nejad adds that Ducktail is the first Facebook-centric malware operation he's aware of that attempts to directly hijack Facebook Business accounts. However, Nejad notes that an earlier Facebook malware operation, dubbed SilentFade, used similar tactics, such as utilizing infostealer logic that leverages Meta's GraphAPI to gather private information about the victims' Facebook account. SilentFade was focused on committing ad fraud.

"However, SilentFade and Ducktail also differ in a few notable ways," Nejad says. "Whereas SilentFade infects victim systems via modified pirated software and potentially unwanted programs, we've observed the Ducktail operation utilizing spear-phishing over a combination of LinkedIn and file/cloud hosting services, in a targeted manner."

And while the SilentFade operation was attributed to a group in China, Nejad says WithSecure has attributed this operation, with high confidence, to an outfit in Vietnam.

Nejad points out the threat actor has continued to update the malware to improve its ability to bypass existing or new Facebook security features alongside other implemented features.

"For instance, one of the latest mechanisms added to the malware allows the threat actor to send a list of email addresses, through their command-and-communication channel, that they would like to use to hijack a specific business," he explains.

**Facebook Business Offers Hackers a Prime Opportunity**

Facebook remains one of the most popular social-network platforms, with close to 3 billion monthly active users, according to its latest quarterly results. That large user base and the wide outreach it provides makes it a perfect platform for advertisers and businesses to operate on — and so, Facebook is one of phishers' favorite brands, according to a recent report.

Just last month, a social-engineering campaign bent on stealing Facebook account credentials and victim phone numbers targeted business pages via a savvy campaign incorporating Facebook's Messenger chatbot feature.

As Ducktail hijacks Facebook business accounts by gaining administrator-level access, it essentially gives the threat actor the ability to gain unlimited access to use the hijacked business account as they wish. This could include carrying out malicious advertising (malvertising), classic fraud efforts (running scams), or to spread disinformation. The threat actor could also potentially use its newfound access to blackmail a company by locking them out of its own business account.

"However, we believe the Ducktail operation uses hijacked business accounts purely to make money by pushing out ads, similar to the SilentFade campaign," Nejad says.

**Review Users, Revoke Access**

Nejad added that to protect themselves from these types of attacks, organizations must exercise caution, practice vigilance, and follow common cybersecurity practices.

"If you believe you've been a victim, we also recommend reviewing users who have been added to your Facebook Business account through Meta's Business Manager, and revoking access for unknown users that were granted Admin access with finance editor role, as well as terminating all browser authentication sessions and resetting your existing login credentials," Nejad says.

Ducktail Spear-Phishing Campaign Uses LinkedIn to Hijack Facebook Business Accounts

Machine learning should be considered an extension of — not a replacement for — existing security methods, systems, and teams.

Contrary to what you may have read, machine learning (ML) isn't magic pixie dust. In general, ML is good for narrowly scoped problems with huge datasets available, and where the patterns of interest are highly repeatable or predictable. Most security problems neither require nor benefit from ML. Many experts, including the folks at Google, suggest that when solving a complex problem you should exhaust all other approaches before trying ML.

ML is a broad collection of statistical techniques that allows us to train a computer to estimate an answer to a question even when we haven't explicitly coded the correct answer. A well-designed ML system applied to the right type of problem can unlock insights that would not have been attainable otherwise.

A successful ML example is natural language processing (NLP). NLP allows computers to "understand" human language, including things like idioms and metaphors. In many ways, cybersecurity faces the same challenges as language processing. Attackers may not use idioms, but many techniques are analogous to homonyms, words that have the same spelling or pronunciations but different meanings. Some attacker techniques likewise closely resemble actions a system administrator might take for perfectly benign reasons.

IT environments vary across organizations in purpose, architecture, prioritization, and risk tolerance. It's impossible to create algorithms, ML or otherwise, that broadly address security use cases in all scenarios. This is why most successful applications of ML in security combine multiple methods to address a very specific issue. Good examples include spam filters, DDoS or bot mitigation, and malware detection.

**Garbage in, Garbage Out**

The biggest challenge in ML is availability of relevant, usable data to solve your problem. For supervised ML, you need a large, correctly labeled dataset. To build a model that identifies cat photos, for example, you train the model on many photos of cats labeled "cat" and many photos of things that aren't cats labeled "not cat." If you don’t have enough photos or they're poorly labeled, your model won't work well.

In security, a well-known supervised ML use case is signatureless malware detection. Many endpoint protection platform (EPP) vendors use ML to label huge quantities of malicious samples and benign samples, training a model on "what malware looks like." These models can correctly identify evasive mutating malware and other trickery where a file is altered enough to dodge a signature but remains malicious. ML doesn't match the signature. It predicts malice using another feature set and can often catch malware that signature-based methods miss.

However, because ML models are probabilistic, there's a trade-off. ML can catch malware that signatures miss, but it may also miss malware that signatures catch. This is why modern EPP tools use hybrid methods that combine ML and signature-based techniques for optimal coverage.

**Something, Something, False Positives**

Even if the model is well-crafted, ML presents some additional challenges when it comes to interpreting the output, including:

*   **The result is a probability.** The ML model outputs the likelihood of something. If your model is designed to identify cats, you'll get results like "this thing is 80% cat." This uncertainty is an inherent characteristic of ML systems and can make the result difficult to interpret. Is 80% cat enough?
*   **The model can't be tuned**, at least not by the end user. To handle the probabilistic outcomes, a tool might have vendor-set thresholds that collapse them to binary results. For example, the cat-identification model may report that anything >90% "cat" is a cat. Your business’s tolerance for cat-ness may be higher or lower than what the vendor set.
*   **False negatives (FN)**, the failure to detect real evil, are one painful consequence of ML models, especially poorly tuned ones. We dislike false positives (FP) because they waste time. But there is an inherent trade-off between FP and FN rates. ML models are tuned to optimize the trade-off, prioritizing the "best" FP-FN rate balance. However, the "correct" balance varies among organizations, depending on their individual threat and risk assessments. When using ML-based products, you must trust vendors to select the appropriate thresholds for you.
*   **Not enough context for alert triage.** Part of the ML magic is extracting powerful predictive but arbitrary "features" from datasets. Imagine that identifying a cat happened to be highly correlated with the weather. No human would reason this way. But this is the point of ML — to find patterns we couldn't otherwise find and to do so at scale. Yet, even if the reason for the prediction can be exposed to the user, it's often unhelpful in an alert triage or incident response situation. This is because the "features" that ultimately define the ML system's decision are optimized for predictive power, not practical relevance to security analysts.

**Would "Statistics" by Any Other Name Smell as Sweet?**

Beyond the pros and cons of ML, there's one more catch: Not all "ML" is really ML. Statistics gives you some conclusions about your data. ML makes predictions about data you didn't have based on data you did have. Marketers have enthusiastically latched onto "machine learning" and "artificial intelligence" to signal a modern, innovative, advanced technology product of some kind. However, there's often very little regard for whether the tech even uses ML, never mind if ML was the right approach.

**So, Can ML Detect Evil or Not?**

ML can detect evil when "evil" is well-defined and narrowly scoped. It can also detect deviations from expected behavior in highly predictable systems. The more stable the environment, the more likely ML is to correctly identify anomalies. But not every anomaly is malicious, and the operator isn't always equipped with enough context to respond. ML's superpower is not in replacing but in extending the capabilities of existing methods, systems, and teams for optimal coverage and efficiency.

The Beautiful Lies of Machine Learning in Security

Company joins AWS Partner Network to provide customers with industrial cybersecurity solution to ensure reliable electricity and fuel supplies.

*   Siemens Energy joins the AWS Partner Network as a Technology Partner to provide customers with industrial cybersecurity, analytics, and storage solutions
*   Siemens Energy’s AI-driven Managed Detection & Response (MDR) solution is a first-of-a-kind security offering available in AWS Marketplace built for energy and utilities, by an integrated energy technology company focused across the entire energy value chain
*   MDR’s monitoring and detection insights provide scalable protection against disruptive cyberattacks in the energy sector using innovative artificial intelligence

Siemens Energy Inc, one of the world’s leading energy technology companies, announces it is joining the Amazon Web Services (AWS) Partner Network (APN), a global community of partners that leverage programs, expertise, and resources to build, market, and sell customer offerings.

This expanded relationship includes listing Siemens Energy’s Managed Detection and Response (MDR) industrial cyber security solution in AWS Marketplace, a digital catalog that makes it easy for customers to find, compare, and immediately start using the software and services that run on AWS. This MDR security offering in AWS Marketplace is built for energy and utilities, by an integrated energy technology company focused across the entire energy value chain. As energy companies strive to protect physical infrastructure from escalating cyber threats, Siemens Energy’s MDR on AWS reduces the cost and technical barriers to achieving strong, fast, comprehensive cyber defense.

Expanding access to state-of-the-art cybersecurity is essential to ensuring reliable electricity and fuel supplies. Cyber risks to infrastructure are escalating amid the ongoing transition toward renewable and low-carbon energy sources and the accompanying expansion of digital devices in the energy sector. Siemens Energy’s MDR will help customers leverage analytics to secure their infrastructure while reducing costs, improving efficacy, and reducing emissions.

Siemens Energy’s industrial monitoring and detection solution defends critical infrastructure against cyberattacks, helping protect communities around the world from supply chain disruptions. AWS’s capabilities allow MDR’s technology to quickly collect and analyze large volumes of data to monitor for cyber threats, giving energy sector chief information security officers (CISOs) the power to detect and uncover attacks before they execute.

Secure cloud capabilities that can integrate digital applications and leverage sensitive data – such as real-time monitoring and detection – add an important and cost-effective tool to the defensive arsenal for CISOs and industrial security analysts.

“The energy transition relies on seamlessly connecting physical assets with digital technologies to foster innovation, reduce emissions, and improve efficiency, but this future depends on strong cybersecurity across the whole supply chain,” said Leo Simonovich, Vice President and Global Head of Industrial Cyber, Siemens Energy. “Siemens Energy’s AI-driven industrial cyber monitoring and detection platform is purpose-built to help CISOs identify, prevent, and mitigate cyber threats to the energy sector’s digital business model. Energy companies at every scale can now access and integrate advanced cyber threat detection across their operating environment, leveraging AWS’s secure cloud platform to build and defend the foundation of a digital energy ecosystem.”

Siemens Energy Takes Next Step to Protect Critical Infrastructure

The open source fully homomorphic encryption library from Duality Technologies is intended to help developers build their own FHE-enabled applications.

While encryption is not a cure-all to address every security challenge, done right, it is an essential component for securing systems, data, and communications. However, doing encryption right is not easy and requires paying careful attention to how it is implemented.

While there are several well-established methods for encrypting data in storage (at rest) and keeping the data encrypted while moving across the network from one system to another (in transit), that isn’t the case for keeping the data encrypted while being processed by applications (in use). Fully homomorphic encryption (FHE) is one way to work with data stored in the cloud or third-party environments while keeping it encrypted.

Several companies have been experimenting with FHE recently. After completing FHE field trials, IBM has begun offering FHE service on IBM Cloud. IBM offers a FHE toolkit for MacOS, iOS, Linux, and Android. Microsoft’s Simple Encrypted Arithmetic Library (SEAL) is a free and open source cross-platform homomorphic encryption library organizations can use to run computations on encrypted data.

FHE currently is slow and has high overhead. Toward that end, Intel is working with Microsoft and DARPA (Defense Advanced Research Projects) to create an ASIC (a specialized microchip customized for a specific purpose) for FHE to help reduce computational overhead and drive down processing time.

And just last week, Duality Technologies released OpenFHE, an open source fully homomorphic encryption library.

"There are several FHE libraries out there, but they suffer from a usability dilemma," said Vinod Vaikuntanathan, co-founder and chief cryptographer at Duality Technologies, in a release. "FHE open source libraries all work on different platforms, implement different features, and have different APIs."

OpenFHE supports advanced FHE features such as bootstrapping, scheme switching, and multiple hardware acceleration backends using the standard Hardware Abstraction Layer (HAL). The associated compilers and other developer tools help developers integrate the library’s encrypted computing capabilities to create their own FHE-enabled applications.

FHE is considered to be the easiest among privacy technology, and OpenFHE is intended to be a “foundational building block” for conducting computations on encrypted data, says Kurt Rohloff, CTO and co-founder of Duality. An example use case allows data providers to encrypt their data locally, aggregate their encrypted data at a central data hub such as a cloud provider, and then run analyses on the data at the hub. All this is possible by using potentially sensitive or private data that doesn’t need to be decrypted.

OpenFHE is the “culmination of years of work” from multiple teams (PALISADE, HElib, and HEAAN) that have “decided to join forces to build the best library possible,” says Rohloff. PALISADE provides a general architecture for an extensible framework that supports multiple post-quantum FHE schemes in a single library, with the ability to integrate general hardware acceleration technologies, he says. HElib provides advanced capabilities for the BGV protocol, allowing for some of the most advanced designs for the most complicated FHE schemes. And finally, HEAAN provides extensive support for CKKS, the protocol most effective for machine learning (ML) applications run on encrypted data.

OpenFHE Brings New Encryption Tools to Developers

The firmware threat offers ultimate stealth and persistence — and may be distributed via tainted firmware components in a supply chain play, researchers theorize.

A Windows firmware rootkit known as "CosmicStrand" has appeared in the cyberthreat firmament, targeting the Unified Extensible Firmware Interface (UEFI) to achieve stealth and persistence.

UEFI firmware is tasked with booting up Windows computers, including the loading of the operating system. As such, if the firmware is tainted with malicious code, that code will launch before the OS does — making it invisible to most security measures and OS-level defenses.

"This, along the fact that the firmware resides on a chip separate from the hard drive, makes attacks against UEFI firmware exceptionally evasive and persistent," researchers from Kaspersky explained in a posting on Monday. "Regardless of how many times the operating system is reinstalled, the malware will stay on the device."

Once triggered, the code deploys a malicious component inside the Windows OS, after a long execution chain. This component connects to a command-and-control server (C2) and waits for instructions to download additional malicious code snippets, which the malware maps into kernel space and assembles into a shellcode.

One shellcode sample obtained by Kaspersky was used to create a new user on the victim's machine and add it to the local administrators group.

"We can infer from this that shellcodes received from the C2 server might be stagers for attacker-supplied PE executables, and it is very likely that many more exist," according to the writeup.

As the US Department of Homeland Security (DHS) and Department of Commerce said in a March report on firmware threats, rootkits present a tremendous amount of risk.

“Attackers can subvert OS and hypervisor visibility and bypass most security systems, hide, and persist in networks and devices for extended periods of time while conducting attack operations, and inflict irrevocable damage,” the government agencies noted in a joint draft report (PDF).

This particular campaign appears highly targeted to specific individuals in China, with some cases seen in Iran and Vietnam, researchers noted. It's unclear what the ultimate endgame for Cosmic Strand is, but it's likely an espionage play; Kaspersky attributed the campaign to an as-yet-unknown Chinese-speaking advanced persistent threat (APT) with overlaps with the MyKings botnet gang.

**Supply Chain, 'Evil Maid' Concerns**

The researchers know very little about how the rootkit is making it onto peoples' machines. That said, supply chain weakness is a possibility, according to Kaspersky, with "unconfirmed accounts discovered online indicating that some users have received compromised devices while ordering hardware components online."

The modifications were specifically introduced to a specific driver by patching it to redirect to malicious code executed during system startup.

"We assess that the modifications \[to the driver\] may have been performed with an automated patcher," the Kaspersky researchers noted. "If so, it would follow that the attackers had prior access to the victim's computer in order to extract, modify and overwrite the motherboard's firmware. This could be achieved through a precursor malware implant already deployed on the computer or physical access (i.e., an evil maid attack scenario)."

They added that in the attacks, the implant burrowed into Gigabyte and ASUS motherboards specifically, which share the H81 chipset. This offers up another possibility for initial compromise.

"This suggests that a common vulnerability may exist that allowed the attackers to inject their rootkit into the firmware's image," according to the report.

**Circa 2016**

Very notably, CosmicStrand appears to have been used in the wild since the end of 2016, long before UEFI attacks were known to be a thing.

"Despite being recently discovered, the CosmicStrand UEFI firmware rootkit seems to have been being deployed for quite a long time," says Ivan Kwiatkowski, senior security researcher at Global Research and Analysis Team (GReAT) at Kaspersky. "This indicates that some threat actors have had very advanced capabilities that they've managed to keep under the radar since 2017. We are left to wonder what new tools they have created in the meantime that we have yet to discover."

UEFI rootkits are still rarely seen in the wild, thanks to how complex and difficult they are to develop — but they're not mythical, either. The first one ever officially spotted was observed by Qihoo 360 to be used by a China-backed APT in 2017; Kaspersky believes CosmicStrand to be related to that threat, which was called the Spy Shadow Trojan.

Then, ESET discovered one in 2018 being used by Russian state-sponsored actor APT28 (aka Fancy Bear, Sednit, or Sofacy). It was dubbed LoJax because of its underlying code, which was a modified version of Absolute Software’s LoJack recovery software for laptops.

Since then, others have infrequently come to light, such as MosaicRegressor and MoonBounce, which Kaspersky found in 2020 and 2022, respectively.

Kaspersky researchers warned that these types of rootkits continue to provide mysteries and raise questions, and deserve more attention from the analyst community.

"CosmicStrand is a sophisticated UEFI firmware rootkit \[that\] appears to have been used in operation for several years, and yet many mysteries remain," they noted. "How many more implants and C2 servers could still be eluding us? What last-stage payloads are being delivered to the victims? But also, is it really possible that CosmicStrand has reached some of its victims through package 'interdiction'? In any case, the multiple rootkits discovered so far evidence a blind spot in our industry that needs to be addressed sooner rather than later."

The feds agree. The aforementioned DHS-led joint draft report noted that firmware presented "a large and ever-expanding attack surface." They added that firmware security is often overlooked, even though it's one of the stealthiest methods by which an attacker can compromise devices at scale.

Rare 'CosmicStrand' UEFI Rootkit Swings into Cybercrime Orbit

Several threat actors used Amadey Bot previously to steal information and distribute malware such as the GandCrab ransomware and the FlawedAmmy RAT.

A dangerous malware variant called "Amadey Bot" that has been largely dormant for the past two years has surfaced again with new features that make it stealthier, more persistent, and much more dangerous than previous versions — including antivirus bypasses.

Amadey Bot first appeared in 2018 and is primarily designed to steal data from infected systems. However, various threat actors — such as Russia's infamous TA505 advanced persistent threat (APT) group — have also used it to distribute other malicious payloads, including GandCrab ransomware and the FlawedAmmy remote access Trojan (RAT), making it a threat for enterprise organizations.

Previously, threat actors used the Fallout and RIG exploit kits, as well as the AZORult infostealer, to distribute Amadey. But researchers at South Korea's AhnLab recently spotted the new variant being installed on systems via SmokeLoader, a malware dropper that attackers have been using since at least 2011.

**Smoke & Mirrors**

Researchers at AhnLab found that the operators of the new Amadey variant have disguised SmokeLoader in software cracks and fake keys for commercial software that people often use to try and activate pirated software. When users download the malware assuming it is a cracked (pirated) version or a key generator, SmokeLoader injects its malicious payload into the currently running Windows Explorer process (explorer.exe) and then proceeds to download Amadey on the infected system, the researchers at AhnLab discovered.

Once the malware is executed, Amadey lodges itself in the TEMP folder as a startup folder, ensuring the malware will persist even after a system reboot. As an additional persistence measure, Amadey also registers itself as a scheduled task in Task Scheduler, according to AhnLab.

After the malware completes its initial setup processes, it contacts a remote, attacker-controlled command-and-control server (C2) and downloads a plug-in to collect environment information. This includes details such as the computer and username, operating system information, a list of applications on the system, and a list of all anti-malware tools on it. 

The sample of the new Amadey variant that researchers at AhnLab analyzed was also designed to take periodic screenshots of the current screen and send them back in a .JPG format to the attacker controlled C2 server.

**Bypassing AV Protections**

AhnLab found that the malware is configured to look for and bypass antivirus tools from 14 vendors, including Avast, Avira, BitDefender, Kaspersky, Sophos, and Microsoft's Windows Defender.

"The new and improved version of the malware flaunts even more features compared to its predecessor," security vendor Heimdal said in a blog post. This includes features "such as scheduled tasks for persistence, advanced reconnaissance, UAC bypassing, and defense evasion strategies tailored for 14 known antivirus products," it noted.

Once Amadey relays system information to the C2 server, the threat actor knows exactly how to bypass protection for the specific AV tools that might be present on the system. "On top of that, once Amadey gets ahold of your AV’s profile, all future payloads or DLLs will be executed with elevated privileges," Heimdal warned in the blog post. 

**A More Dangerous Version of Amadey**

The information that Amadey relays to the C2 server allows the attackers to take a variety of follow-up actions, including installing additional malware. The sample that AhnLab analyzed, for instance, downloaded a plug-in for stealing Outlook emails and information about FTPs and VPN clients on the infected system. 

It also installs an additional information stealer called RedLine on the victim system. RedLine is a prolific information stealer that first surfaced in 2020 and has been distributed via various mechanisms, including COVID-19 themed phishing emails, fake Google ads and in targeted campaigns. Researchers from Qualys recently observed the malware being distributed via fake cracked software on Discord.  

Researchers from BlackBerry Cylance who analyzed the earlier version of Amadey determined at the time that the malware does not install any additional payloads if it assesses the victim to be in Russia.

Dmitry Bestuzhev, threat researcher at BlackBerry says SmokeLoader has been actively tricking victims via cracked software and so-called “software activation” programs from the beginning of this year. 

"We saw several brands being abused by cybercriminals spreading malicious cracks for the most popular software houses," he says. 

Amadey's botnet too has been very active, impersonating popular private and secure email vendors to infect targets. "According to our telemetry, most of them are in the US, followed by Japan, Mexico and Brazil," Bestuzhev says.

Supercharged Version of Amadey Infostealer & Malware Dropper Bypasses AVs

After leaking 80 million US customer data records in a cyberattack last summer, T-Mobile offers to settle a wide-ranging class action suit for just $350 million.

Names, driver's licenses, and even Social Security numbers of more than 80 million T-Mobile customers in the US were compromised in a cyberattack last summer. Now T-Mobile is offering the victims a settlement in a class-action lawsuit stemming from the breach. It's cold comfort to victims, though — the compensation works out to just over $4 per person.

In a filing with the US Securities and Exchange Commission (SEC), the company said it would pay $350 million to fund customer claims, and pick up the attorneys' fees and administrative costs of handling the payouts. The mobile carrier also pledged to invest an additional $150 million in cybersecurity through 2023.

Altogether the company said the settlement offer would cost T-Mobile $400 million in the second quarter of 2022. To put that in perspective, T-Mobile's US revenue for the 12 months ending March 31 was $80.5 billion, a 4.46% increase year-over-year.

The T-Mobile SEC filing added that the settlement doesn't come with any admission of liability, wrongdoing, or responsibility on the part of the company, and added that T-Mobile expects to receive court approval of the settlement terms by as early as December.

The proposed penalty is in line with similar data breach settlements, like the $190 million settlement Capital One reached for leaking 100 million credit card applications from the US and another 6 million from Canada in a 2019 cyberattack — less than $2 per victim.

As a financial institution, Capital One was also ordered to pay an $80 million fine to Federal Reserve regulators. Both sums are a drop in the bucket for the corporate giant. Capital One's annual net income for 2021 was $12 billion, a 403.79% increase from 2020.

Keep up with the latest cybersecurity threats, newly-discovered vulnerabilities, data breach information, and emerging trends. Delivered daily or weekly right to your email inbox.

Subscribe

T-Mobile Pitches $4-Per-Customer Settlement for Data Leak Impacting 80M People

To help discern legitimate traffic from fraud, it helps to understand user intent as shown through their behavior.

For many years, security monitoring relied on gathering data from layer 4 of the OSI model through such data types as NetFlow. Because layer 4 data dealt with the transport layer, it isn't the most informative — though for a period of time, it was what security teams could reliably get access to and efficiently query. Then, as technology improved, security teams found themselves with access to a much richer data set: layer 7 data. Proxy logs, DNS logs, packet capture (PCAP), and other layer 7 data sources became available, and it was a game-changer for security teams.

Layer 7 data allows us to interrogate the application layer. Specifically, as it relates to digital channels such as Web and mobile, layer 7 data lets us understand what is happening within the end-user application session. This gives us essential context around the end user's activity. Unfortunately, layer 7 data does not allow us to understand the "how" behind what is happening. Questions such as "How is the end user behaving?", "What is the end user's intent?", and "Is this legitimate end-user activity?" can only be answered by looking beyond layer 7.

To understand intent — the "how" behind the "what" — we need to closely examine the behavior of the end user in the session. This additional behavioral insight is critical to an enterprise's ability to separate legitimate traffic from fraud. In other words, the difference between the legitimate use of an application and abuse of that application (i.e., fraud) is the intent of the end user responsible for the activity. When we look at the concept of fraud in this manner, it is easy to see that visibility into "what" the end user is doing inside the application session isn't enough. We also need visibility into "how" they are doing it.

**Behaviors That Could Signal Fraudulent Use**

Some people refer to this end-user layer above layer 7 of the OSI model as layer 8. And as the Sesame Street song says, eight is great. Let's take a look at some of the ways in which layer 8 data can help us better detect fraud.

**Optimized mouse movements.** Legitimate users tend to have very random mouse movements when interacting with an application. The reason is simple: Legitimate users aren't interacting with the application "professionally" and thus don't have any need or incentive to optimize their mouse movements. Fraudsters, on the other hand, who may be trying to access tens, hundreds, or thousands of accounts fraudulently, have every motivation to optimize their mouse movements to save time.

**Pasting.** I don't know about you, but I don't often cut and paste my username and password or first name and last name from a text file. As it turns out, most legitimate users don't either. Fraudsters, as you might imagine, do this quite frequently, particularly when it comes to account takeover (ATO).

**Strange keys.** If you are a legitimate user, chances are that you use a fairly standard set of letters, numbers, and special characters when interacting with an application. It is fairly unlikely that you would use function keys, keyboard shortcuts, or other unusual combinations. Fraudsters who are looking to save time, however, often do exactly that.

**A signature device.** Fraudsters typically have one or a few favorite devices that they have configured exactly as they want them. Fraudsters will often use these same devices to log in to a relatively large number of accounts on the same application. Because of this, if we invest in accurate and reliable device identification and track logins by device, we can often use that knowledge to understand when we might be dealing with a fraudulent session.

**Other tricks.** Fraudsters often rely on environment spoofing, VPN, and other tricks to try to appear to be legitimate users. Legitimate users do this far less frequently, though it does still happen.

The above user behaviors are a few examples of the differences in behavior between legitimate users and fraudsters. None of these behaviors in and of themselves can tell us with 100% certainty whether a given session is legitimate or fraudulent. They can, however, provide us valuable insight into the "how" behind the "what." That, in turn, can help us make far more accurate assessments around what is fraud. Understanding end-user behavior (layer 8 data) allows us to increase our detection rates, while at the same time lowering our false positive rates.

Source

DARKReading