Blog
/
AI
/
February 10, 2025

From Hype to Reality: How AI is Transforming Cybersecurity Practices

Default blog imageDefault blog imageDefault blog imageDefault blog imageDefault blog imageDefault blog image
10
Feb 2025
AI hype is everywhere, but not many vendors are getting specific. Darktrace’s multi-layered AI combines various machine learning techniques for behavioral analytics, real-time threat detection, investigation, and autonomous response.

AI is everywhere, predominantly because it has changed the way humans interact with data. AI is a powerful tool for data analytics, predictions, and recommendations, but accuracy, safety, and security are paramount for operationalization.

In cybersecurity, AI-powered solutions are becoming increasingly necessary to keep up with modern business complexity and this new age of cyber-threat, marked by attacker innovation, use of AI, speed, and scale. The emergence of these new threats calls for a varied and layered approach in AI security technology to anticipate asymmetric threats.

While many cybersecurity vendors are adding AI to their products, they are not always communicating the capabilities or data used clearly. This is especially the case with Large Language Models (LLMs). Many products are adding interactive and generative capabilities which do not necessarily increase the efficacy of detection and response but rather are aligned with enhancing the analyst and security team experience and data retrieval.

Consequently, many  people erroneously conflate generative AI with other types of AI. Similarly, only 31% of security professionals report that they are “very familiar” with supervised machine learning, the type of AI most often applied in today’s cybersecurity solutions to identify threats using attack artifacts and facilitate automated responses. This confusion around AI and its capabilities can result in suboptimal cybersecurity measures, overfitting, inaccuracies due to ineffective methods/data, inefficient use of resources, and heightened exposure to advanced cyber threats.

Vendors must cut through the AI market and demystify the technology in their products for safe, secure, and accurate adoption. To that end, let’s discuss common AI techniques in cybersecurity as well as how Darktrace applies them.

Modernizing cybersecurity with AI

Machine learning has presented a significant opportunity to the cybersecurity industry, and many vendors have been using it for years. Despite the high potential benefit of applying machine learning to cybersecurity, not every AI tool or machine learning model is equally effective due to its technique, application, and data it was trained on.

Supervised machine learning and cybersecurity

Supervised machine models are trained on labeled, structured data to facilitate automation of a human-led trained tasks. Some cybersecurity vendors have been experimenting with supervised machine learning for years, with most automating threat detection based on reported attack data using big data science, shared cyber-threat intelligence, known or reported attack behavior, and classifiers.

In the last several years, however, more vendors have expanded into the behavior analytics and anomaly detection side. In many applications, this method separates the learning, when the behavioral profile is created (baselining), from the subsequent anomaly detection. As such, it does not learn continuously and requires periodic updating and re-training to try to stay up to date with dynamic business operations and new attack techniques. Unfortunately, this opens the door for a high rate of daily false positives and false negatives.

Unsupervised machine learning and cybersecurity

Unlike supervised approaches, unsupervised machine learning does not require labeled training data or human-led training. Instead, it independently analyzes data to detect compelling patterns without relying on knowledge of past threats. This removes the dependency of human input or involvement to guide learning.

However, it is constrained by input parameters, requiring a thoughtful consideration of technique and feature selection to ensure the accuracy of the outputs. Additionally, while it can discover patterns in data as they are anomaly-focused, some of those patterns may be irrelevant and distracting.

When using models for behavior analytics and anomaly detection, the outputs come in the form of anomalies rather than classified threats, requiring additional modeling for threat behavior context and prioritization. Anomaly detection performed in isolation can render resource-wasting false positives.

LLMs and cybersecurity

LLMs are a major aspect of mainstream generative AI, and they can be used in both supervised and unsupervised ways. They are pre-trained on massive volumes of data and can be applied to human language, machine language, and more.

With the recent explosion of LLMs in the market, many vendors are rushing to add generative AI to their products, using it for chatbots, Retrieval-Augmented Generation (RAG) systems, agents, and embeddings. Generative AI in cybersecurity can optimize data retrieval for defenders, summarize reporting, or emulate sophisticated phishing attacks for preventative security.

But, since this is semantic analysis, LLMs can struggle with the reasoning necessary for security analysis and detection consistently. If not applied responsibly, generative AI can cause confusion by “hallucinating,” meaning referencing invented data, without additional post-processing to decrease the impact or by providing conflicting responses due to confirmation bias in the prompts written by different security team members.

Combining techniques in a multi-layered AI approach

Each type of machine learning technique has its own set of strengths and weaknesses, so a multi-layered, multi-method approach is ideal to enhance functionality while overcoming the shortcomings of any one method.

Darktrace’s multi-layered AI engine is powered by multiple machine learning approaches, which operate in combination for cyber defense. This allows Darktrace to protect the entire digital estates of the organizations it secures, including corporate networks, cloud computing services, SaaS applications, IoT, Industrial Control Systems (ICS), and email systems.

Plugged into the organization’s infrastructure and services, our AI engine ingests and analyzes the raw data and its interactions within the environment and forms an understanding of the normal behavior, right down to the granular details of specific users and devices. The system continually revises its understanding about what is normal based on evolving evidence, continuously learning as opposed to baselining techniques.

This dynamic understanding of normal partnered with dozens of anomaly detection models means that the AI engine can identify, with a high degree of precision, events or behaviors that are both anomalous and unlikely to be benign. Understanding anomalies through the lens of many models as well as autonomously fine-tuning the models’ performances gives us a higher understanding and confidence in anomaly detection.

The next layer provides event correlation and threat behavior context to understand the risk level of an anomalous event(s). Every anomalous event is investigated by Cyber AI Analyst that uses a combination of unsupervised machine learning models to analyze logs with supervised machine learning trained on how to investigate. This provides anomaly and risk context along with investigation outcomes with explainability.

The ability to identify activity that represents the first footprints of an attacker, without any prior knowledge or intelligence, lies at the heart of the AI system’s efficacy in keeping pace with threat actor innovations and changes in tactics and techniques. It helps the human team detect subtle indicators that can be hard to spot amid the immense noise of legitimate, day-to-day digital interactions. This enables advanced threat detection with full domain visibility.

Digging deeper into AI: Mapping specific machine learning techniques to cybersecurity functions

Visibility and control are vital for the practical adoption of AI solutions, as it builds trust between human security teams and their AI tools. That is why we want to share some specific applications of AI across our solutions, moving beyond hype and buzzwords to provide grounded, technical explanations.

Darktrace’s technology helps security teams cover every stage of the incident lifecycle with a range of comprehensive analysis and autonomous investigation and response capabilities.

  1. Behavioral prediction: Our AI understands your unique organization by learning normal patterns of life. It accomplishes this with multiple clustering algorithms, anomaly detection models, Bayesian meta-classifier for autonomous fine-tuning, graph theory, and more.
  2. Real-time threat detection: With a true understanding of normal, our AI engine connects anomalous events to risky behavior using probabilistic models. 
  3. Investigation: Darktrace performs in-depth analysis and investigation of anomalies, in particular automating Level 1 of a SOC team and augmenting the rest of the SOC team through prioritization for human-led investigations. Some of these methods include supervised and unsupervised machine learning models, semantic analysis models, and graph theory.
  4. Response: Darktrace calculates the proportional action to take in order to neutralize in-progress attacks at machine speed. As a result, organizations are protected 24/7, even when the human team is out of the office. Through understanding the normal pattern of life of an asset or peer group, the autonomous response engine can isolate the anomalous/risky behavior and surgically block. The autonomous response engine also has the capability to enforce the peer group’s pattern of life when rare and risky behavior continues.
  5. Customizable model editor: This layer of customizable logic models tailors our AI’s processing to give security teams more visibility as well as the opportunity to adapt outputs, therefore increasing explainability, interpretability, control, and the ability to modify the operationalization of the AI output with auditing.

See the complete AI architecture in the paper “The AI Arsenal: Understanding the Tools Shaping Cybersecurity.”

Figure 1. Alerts can be customized in the model editor in many ways like editing the thresholds for rarity and unusualness scores above.

Machine learning is the fundamental ally in cyber defense

Traditional security methods, even those that use a small subset of machine learning, are no longer sufficient, as these tools can neither keep up with all possible attack vectors nor respond fast enough to the variety of machine-speed attacks, given their complexity compared to known and expected patterns.

Security teams require advanced detection capabilities, using multiple machine learning techniques to understand the environment, filter the noise, and take action where threats are identified.

Darktrace’s multi-layered AI comes together to achieve behavioral prediction, real-time threat detection and response, and incident investigation, all while empowering your security team with visibility and control.

Download the full report

Discover specifically how Darktrace applies different types of AI to improve cybersecurity efficacy and operations in this technical paper.

Inside the SOC
Darktrace cyber analysts are world-class experts in threat intelligence, threat hunting and incident response, and provide 24/7 SOC support to thousands of Darktrace customers around the globe. Inside the SOC is exclusively authored by these experts, providing analysis of cyber incidents and threat trends, based on real-world experience in the field.
Author
Nicole Carignan
SVP, Security & AI Strategy, Field CISO

Nicole Carignan is Senior Vice President, Security and AI Strategy, Field CISO at Darktrace. In this role, Nicole is focused on providing technical, strategic guidance and expertise in cybersecurity, threat research, AI and data science to customers, partners, and organizations within Darktrace. An expert in the safe, secure and responsible application of AI in cybersecurity, Nicole engages in product innovation and advisory, thought leadership, research, and cybersecurity and AI community engagement to ensure Darktrace delivers solutions that meet customer’s evolving needs. Her insights have been cited by global cybersecurity publications, and she is a frequent speaker at industry

conferences and professional associations such as Women in Data Science, AUSA Cyberworld, ISACA Houston and more.

With over 25 years of experience, Nicole has deep expertise in data science, machine learning, cybersecurity, threat intelligence, operations and network engineering. Prior to Darktrace, Nicole served as the Head of North America Operations for CounterCraft Security, a leader in deception technology and threat intelligence solutions. Nicole worked in and supported the US federal government for over 20 years, serving in multiple technical and operational roles for the Intelligence Community and US Department of Defense and consulted on multiple large scale data science efforts.

Book a 1-1 meeting with one of our experts
Share this article

More in this series

No items found.

Blog

/

AI

/

March 25, 2025

Survey Findings: AI Cybersecurity Priorities and Objectives in 2025

Default blog imageDefault blog image

AI is changing the cybersecurity field, both on the offensive and defensive sides. We surveyed over 1,500 cybersecurity professionals from around the world to uncover their attitudes, understanding, and priorities when it comes to AI cybersecurity in 2025. Our full report, unearthing some telling trends, is available now.  

Download the full report to explore these findings in depth

It is clear that security professionals know their field is changing fast, and that AI will continue to influence those changes. Our survey results show that they are aware that the rise of AI will require them to adopt new tools and learn to use them effectively. Still, they aren’t always certain about how to plan for the future, or what to invest in.

The top priorities of security stakeholders for improving their defenses against AI-powered threats include augmenting their existing tool stacks with AI-powered solutions and improving integration among their security tools.

Figure 1: Year-over-year changes to the priorities of securitystakeholders.

Increasing cybersecurity staff

As was also the case last year, security stakeholders are less interested in hiring additional staff than in adding new AI-powered tools onto their existing security stacks, with only with 11% (and only 8% of executives) planning to increase cybersecurity staff in 2025.

This suggests that leaders are looking for new methods to overcome talent resource shortages.

Adding AI-powered security tools to supplement existing solutions

Executives are particularly enthusiastic about adopting AI-driven tools. Within that goal, there is consensus about the qualities cyber professionals are looking for when purchasing new security capabilities or replacing existing products.

  • 87% of survey respondents prefer solutions that are part of a broader platform over individual point products

These results are similar to last year’s, where again, almost nine out of ten agreed that a platform-oriented security solution was more effective at stopping cyber threats than a collection of individual products.

  • 88% of survey respondents agree that the use of AI within the security stack is critical to freeing up time for security teams to become more proactive, compared to reactive

AI itself can contribute to this shift from reactive to proactive security, improving risk prioritization and automating preventative strategies like Attack Surface Management (ASM) and proactive exposure management.

  • 84% of survey respondents prefer defensive AI solutions that do not require the organization’s data to be shared externally

This preference may reflect increasing attention to the data privacy and security risks posed by generative AI (gen AI) adoption. It may also reflect growing awareness of data residency requirements and other restrictions that regulators are imposing.

Improving cybersecurity awareness training for end users

Based on the survey results, practitioners in SecOps are more interested in improving security awareness training.

This goal is not necessarily mutually exclusive from the addition of AI tools. For example, teams can leverage AI to build more effective security awareness training programs, and as gen AI tools are adopted, users will need to be taught about data privacy and associated security risks.

Looking towards the future

One conclusion we can draw from the attitudinal shifts from last year’s survey to this year’s: while hiring more security staff might be a nice-to-have, implementing AI-powered tools so that existing employees can work smarter is increasingly viewed as a must-have.

However, trending goals are not just about managing resources, whether headcount or AI investments, to keep up with workloads. Existing end users must also be trained to follow safe practices while using established and newly adopted tools.

Security professionals, including executives, SecOps, and every role in between, continue to shift their identified challenges and priorities as they gear up for the coming year in the Era of AI.

State of AI report

Download the full report to explore these findings in depth

The full report for Darktrace’s State of AI Cybersecurity is out now. Download the paper to dig deeper into these trends, and see how results differ by industry, region, organization size, and job title.  

Continue reading
About the author
The Darktrace Community

Blog

/

Network

/

March 21, 2025

Cyberhaven Supply Chain Attack: Exploiting Browser Extensions

Default blog imageDefault blog image

The evolution of supply chain attacks

Supply chain attacks are becoming increasingly sophisticated. As network defenses improve, threat actors continuously adapt and refine their tactics, techniques, and procedures (TTPs) to achieve their goals. In recent years, this has led to a rise in the exploitation of trusted services and software, including legitimate browser extensions. Exploitation of these extensions can provide adversaries with a stealthy means to infiltrate target networks and access high-value accounts undetected.

A notable example of this trend was the compromise of the Cyberhaven Chrome extension at the end of 2024. This incident appeared to be part of a broader campaign targeting multiple Chrome browser extensions, highlighting the evolving nature of supply chain attacks [1].

What is Cyberhaven?

Cyberhaven, a US-based data security organization, experienced a security breach on December 24, 2024, when a phishing attack reportedly compromised one of their employee's credentials [2]. This allowed attackers to publish a malicious version of the Cyberhaven Chrome extension, which exfiltrated cookies and authenticated sessions from targeted websites. The malicious extension was active from December 25 to December 26 – a time when most businesses and employees were out of office and enjoying the festive period, a fact not lost on threat actors. The attackers, likely a well-organized and financially motivated group, compromised more than 30 additional Chrome extensions, affecting more than 2.6 million users [3]. They used sophisticated phishing techniques to authorize malicious OAuth applications, bypassing traditional security measures and exploiting vulnerabilities in OAuth authorizations. The primary motive appeared to be financial gain, targeting high-value platforms like social media advertising and AI services [4].

In late December 2024, multiple Darktrace customers were compromised via the Cyberhaven Chrome extension; this blog will primarily focus on Darktrace / NETWORK detections from one affected customer.

Darktrace’s coverage of Cyberhaven compromises

On December 26, 2024, Darktrace identified a series of suspicious activities across multiple customer environments, uncovering a structured attack sequence that progressed from initial intrusion to privilege escalation and data exfiltration. The attack was distributed through a malicious update to the Cyberhaven Chrome extension [2]. The malicious update established a foothold in customer environments almost immediately, leading to further anomalies.

As with other Chrome browser extensions, Cyberhaven Chrome extensions were updated automatically with no user interaction required. However, in this instance, the automatic update included a malicious version which was deployed to customer environments. This almost immediately introduced unauthorized activity, allowing attackers to establish a foothold in customer networks. The update allowed attackers to execute their objectives in the background, undetected by traditional security tools that rely on known indicators of compromise (IoCS) rather than identifying anomalies.

While multiple customer devices were seen connecting to cyberhaven[.]io, a legitimate Cyberhaven domain, Darktrace detected persistent beaconing behavior to cyberhavenext[.]pro, which appeared to be attempting to masquerade as another legitimate Cyberhaven domain. Darktrace recognized this activity as unusual, triggering several model alerts in Darktrace / NETWORK to highlight the persistent outbound connections to the suspicious domain.

Further analysis of external connectivity patterns indicated  an increase in anomalous HTTP requests alongside this beaconing activity. Multiple open-source intelligence (OSINT) sources also suggest that the cyberhavenext[.]pro endpoint is associated with malicious activities [5].

Darktrace / NETWORK’s detection of beaconing activity to cyberhavenext[.]pro
Figure 1: Darktrace / NETWORK’s detection of beaconing activity to cyberhavenext[.]pro

Analysis using Darktrace’s Advanced Search revealed that some of these connections were directed to the suspicious external IP address 149.28.124[.]84. Further investigation confirmed that the IP correlated with two SSL hostnames, including the malicious cyberhavenext[.]pro, further reinforcing its connection to the attack infrastructure.

Darktrace Advanced Search analysis showing the IP address 149.28.124[.]84 correlating to two SSL hostnames, one of which is cyberhavenext[.]pro.
Figure 2: Darktrace Advanced Search analysis showing the IP address 149.28.124[.]84 correlating to two SSL hostnames, one of which is cyberhavenext[.]pro.

Between December 23 and December 27, Darktrace observed sustained beaconing-like activity from affected devices on the customer’s network.

Darktrace’s detection of beaconing activities from a customer device to the endpoint 149.28.124[.]84 between December 23 and December 27.
Figure 3: Darktrace’s detection of beaconing activities from a customer device to the endpoint 149.28.124[.]84 between December 23 and December 27.

Darktrace observed 27 unique devices connecting to the malicious command-and-control (C2) infrastructure as far back as December 3. While most connections were brief, they represented an entry point for malicious activity. Over a two-day period, two devices transmitted 5.57 GiB of incoming data and 859.37 MiB of outgoing data, generating over 3 million log events across SSL, HTTP, and connection data.

Subsequent analysis identified a significant increase in unauthorized data transfers to the aforementioned 149.28.124[.]84 IP on another customer network, highlighting the potential broader impact of this compromise. The volume and frequency of these transfers suggested that attackers were leveraging automated data collection techniques, further underscoring the sophistication of the attack.

Darktrace’s detection of the likely exfiltration of 859.37 MiB to the endpoint 149.28.124[.]84.
Figure 4: Darktrace’s detection of the likely exfiltration of 859.37 MiB to the endpoint 149.28.124[.]84.

External research suggested that once active, the Cyberhaven extension would begin silently collecting session cookies and authentication tokens, specifically targeting high-value accounts such as Facebook Ads accounts [4]. Darktrace’s analysis of another affected customer noted many HTTP POST connections directed to a specific URI ("ai-cyberhaven"), while GET requests contained varying URIs prefixed with "/php/urlblock?args=AAAh....--redirect." This activity indicated an exfiltration mechanism, consistent with techniques observed in other compromised Chrome extensions. By compromising session cookies, attackers could potentially gain administrative access to connected accounts, further escalating their privileges [4].

Conclusion

This incident highlights the importance of monitoring not just endpoint security, but also cloud and browser-based security solutions, as attackers increasingly target these trusted and oft overlooked vectors.

Ultimately, by focusing on anomaly detection and behavioral analysis rather than static signatures and lists of ‘known bads’, Darktrace was able to successfully detect devices affected by the Cyberhaven Chrome browser extension compromise, by identifying activity that would likely have been considered legitimate and benign by traditional security solutions.

This compromise also serves as a reminder that supply chain attacks are not limited to traditional software vendors. Browser extensions, cloud-based applications, and SaaS services are equally vulnerable, as evidenced by Darktrace's detection of Balada Injector malware exploiting WordPress vulnerabilities to gain unauthorized network access [6]. Therefore, increased targeting of browser-based security tools, and a greater exploitation of OAuth and session hijacking techniques are to be expected. Attackers will undoubtedly refine their methods to infiltrate legitimate vendors and distribute malicious updates through trusted channels. By staying informed, vigilant, and proactive, organizations can mitigate exposure to evolving supply chain threats and safeguard their critical assets from emerging browser-based attack techniques.

Credit to Rajendra Rushanth (Cyber Analyst) Justin Torres (Senior Cyber Analyst) and Ryan Traill (Analyst Content Lead)

Appendices

Darktrace Model Detections

·       Compromise / Beaconing Activity To External Rare (AP: C2 Comms)

·       Compromise / Beacon for 4 Days (AP: C2 Comms)

·       Compromise / HTTP Beaconing to Rare Destination (AP: C2 Comms)

·       Device / Suspicious Domain (AP: C2 Comms, AP: Tooling)

·       Compromise / Sustained TCP Beaconing Activity To Rare Endpoint (AP: C2 Comms)

·       Anomalous Server Activity / Rare External from Server (AP: C2 Comms)

·       Anomalous Connection / Multiple Failed Connections to Rare Endpoint (AP: C2 Comms)

·       Anomalous Server Activity / Anomalous External Activity from Critical Network Device (AP: C2 Comms)

·       Compromise / Slow Beaconing Activity To External Rare (AP: C2 Comms)

·       Compromise / Repeating Connections Over 4 Days (AP: C2 Comms)

·       Anomalous Connection / Multiple HTTP POSTs to Rare Hostname (AP: C2 Comms)

·       Anomalous Server Activity / Outgoing from Server (AP: C2 Comms)

·       Compromise / High Volume of Connections with Beacon Score (AP: C2 Comms)

·       Compromise / Large Number of Suspicious Failed Connections (AP: C2 Comms)

·       Email Nexus / Connection to Hijacked Correspondent Link

·       Compromise / Suspicious TLS Beaconing To Rare External (AP: C2 Comms)

·       Compromise / Quick and Regular Windows HTTP Beaconing (AP: C2 Comms)

List of IoCs

IoC - Type - Description + Confidence

cyberhavenext[.]pro - Hostname - Used for C2 communications and data exfiltration (cookies and session tokens)

149.28.124[.]84 - IP - Associated with malicious infrastructure

45.76.225[.]148 - IP - Associated with malicious infrastructure

136.244.115[.]219 - IP - Associated with malicious infrastructure

MITRE ATT&CK Mapping

Tactic – Technique – Sub-Technique

INITIAL ACCESS - T1176 - Browser Extensions

EXECUTION - T1204.002 - Malicious Browser Extensions

PERSISTENCE - T1176 - Browser Extensions

COMMAND AND CONTROL - T1071.001 - Web Protocols

COMMAND AND CONTROL - T1001 - Data Obfuscation

CREDENTIAL ACCESS - T1539 - Steal Web Session Cookie

DISCOVERY - T1518.001 - Security Software Discovery

LATERAL MOVEMENT - T1557.003 - Man-in-the-Browser

EXFILTRATION - T1041 - Exfiltration Over C2 Channel

EXFILTRATION - T1567.002 - Exfiltration to Cloud Storage

IMPACT - T1583.006 - Session Hijacking

References

[1] https://thehackernews.com/2024/12/16-chrome-extensions-hacked-exposing.html

[2] https://www.cyberhaven.com/blog/cyberhavens-chrome-extension-security-incident-and-what-were-doing-about-it

[3] https://www.infosecurity-magazine.com/news/chrome-browser-extensions-hijacked/

[4] https://www.theverge.com/2024/12/28/24330758/chrome-extension-cyberhaven-hijack-phishing-cyberattack-facebook-ads-authentication-theft

[5] https://www.virustotal.com/gui/domain/cyberhavenext.pro

[6] https://darktrace.com/blog/balada-injector-darktraces-investigation-into-the-malware-exploiting-wordpress-vulnerabilities

Continue reading
About the author
Rajendra Rushanth
Cyber Analyst
Your data. Our AI.
Elevate your network security with Darktrace AI