How to Prevent Spear Phishing Attacks Post Twitter Hack
Twitter confirmed spear phishing as the cause of last month's attack. Learn about the limits of current defenses against spear phishing and how AI can stop it.
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.
Written by
Dan Fein
VP, Product
Share
04
Aug 2020
Twitter has now confirmed that it was a “phone spear phishing attack” targeting a small number of their employees that allowed hackers to access 130 high-profile user accounts and fool thousands of people into giving away money via bitcoin.
Spear phishing involves targeted texts or emails aimed at individuals in an attempt to ‘hook’ them into opening an attachment or malicious link. This attack highlights the limitations in the security controls adopted by even some of the largest and most tech-savvy organizations out there, who continue to fall victim to this well-known attack technique.
The incident has been described by Twitter as a “coordinated social engineering attack” that “successfully targeted employees with access to internal systems and tools.”
Though the specific nature of the attack remains unclear, it likely followed a similar pattern to the series of threat finds detailed elsewhere on the Darktrace Blog: impersonating trusted colleagues or platforms, such as WeTransfer, Microsoft Teams or even Twitter itself, with an urgent message coaxing an employee into clicking on a disguised URL and inputting their credentials on a fake login page.
When an employee inputs their credentials, that data is recorded and beaconed back to the attacker, who will then use these login details to access internal systems — which, in this case, allowed them to subsequently take control of celebrities’ Twitter accounts and send out the damaging Tweets that left thousands out of pocket.
Training the workforce is not enough
Twitter says in a statement that this incident has forced them to “accelerate several of [their] pre-existing security workstreams.” But the suggestion that they will continue to organize “ongoing company-wide phishing exercises throughout the year” indicates an over-reliance on the ability of humans to identify these malicious email attacks that are getting more and more advanced, and harder to distinguish from genuine communication.
Cyber-criminals are now using AI to create fake profiles, personalize messages and replicate communication patterns, at a speed and scale that no human ever could. In this threat landscape, there can no longer be a reliance solely on educating the workforce, as the difference between a malicious email and legitimate communication becomes almost imperceptible. This has led to an acceptance that we must rely on technology to help us catch the subtle signs of attack, when humans alone fail to do so.
The legacy approach: no playbook for new attacks
The majority of communications security systems are not where they need to be, and this is particularly true for the email realm. Most tools in use today rely on static blacklists of rules and signatures that analyze emails in isolation, against known ‘bads’. Methods like looking for IP addresses or file hashes associated with phishing have had limited success in stopping attackers, who have devised simple techniques to bypass them.
The mean lifetime of an attack has reduced from 2.1 days in 2018 to 0.5 days in 2020. As soon as an SEG identifies a domain or a file hash as malicious, cyber-criminals change their attack infrastructure and launch a new wave of fresh attacks. Their fundamental means of operation renders legacy security tools incapable of evolving with the threat landscape, and it is for this reason that over 94% of cyber-attacks today start with an email.
How Cyber AI catches the threats others miss
However, one area where email security has seen great progress even in the last two years is the application of AI to spot the subtle features of advanced email attacks, even those that leverage novel malware. This approach allows security tools to move away from the binary decision-making that comes with asking “Is this email ‘bad’?” and moving to the far more useful question of “does this belong?”
This form of what we’re calling ‘layered AI’ combines supervised and unsupervised machine learning, enabling it to spot the subtle deviations from learned ‘patterns of life’ that are indicative of a cyber-threat.
Supervised machine learning models can be trained on millions of emails to find subtle patterns undetectable by humans and detect new variations of known threat types. These models are able to find the real-world intentions behind an email: by training on millions of spear phishing emails, for example, a system can find patterns associated with this type of email attack and accurately classify a future email as spear phishing.
In addition, unsupervised machine learning models can be trained on all available email data for an organization to find unknown variations of unknown threat types — that is, the ‘unknown unknowns,’ the combinations never before seen. Ultimately this is what enables a system to ask that critical question “does this belong?” and spot genuine anomalies that fall outside of the norm.
Layering both of these applications of AI allows us to make determinations such as: ‘this is a phishing email and it doesn’t belong’, dramatically improving the system’s accuracy and allowing it to interrupt only the malicious emails – since there could be phishy-looking emails that are legitimate! It also enables us to act in proportion to the threat identified: locking links and attachments in some cases, or holding back emails entirely in others.
This form of ‘layered AI’ requires an advanced understanding of mathematics and machine learning that takes years of research and development. With that experience, Cyber AI has proven itself capable of catching the full range of advanced attacks targeting the inbox, from spear phishing and impersonation attempts, to account takeovers and supply chain attacks. Once implemented, it takes only a week before any new organization can derive value, and thousands of customers now rely on Cyber AI to protect both their email realm and wider network.
Plenty more phish in the sea
This will not be the last time this year that a cyber-attack caused by spear phishing makes the headlines. Just this week, it was revealed that Russian-backed cyber-criminals stole sensitive documents on US-UK trade talks after successful spear phishing, and the technique may well have played a part in ongoing vaccine research espionage that surfaced in July.
With the US presidential race heating up, it was recently revealed that fewer than 3 out of 10 election administrators have basic controls to prevent phishing. This attack method may come to not only damage organizations and their reputation, but also to undermine the trust that serves as the bedrock of democracy. Now is the time to start recognizing the very real threat that email attackers represent, and to prepare our defenses accordingly.
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.
Hola VPN Abuse: From Proxy Traffic to Malware and Cryptomining
Introduction
In enterprise environments, non-compliant software traffic can introduce unexpected exposure by creating unmanaged paths for outbound connectivity. Hola VPN is a notable example because of its peer-to-peer design, which can effectively turn user devices into routing or exit nodes for other parties’ traffic, shifting the risk profile from that of a traditional virtual private network (VPN) to something closer to a distributed proxy.
As a result, the appearance of Hola-related activity, whether from prior installation or unintended background connections, should be treated with caution. Such activity may provide a foothold for malicious behavior, including lateral movement or command-and-control communication.
This blog explores how Hola-associated activity appeared as part of broader patterns of suspicious behavior observed across the Darktrace customer base.
The campaign
In February and March 2026, Darktrace observed similar anomalous activity across multiple customer environments, with affected devices showing consistent behavioral patterns. These included connections to multiple *.hola[.]org endpoints using Hola-related user agents, suggesting interaction with Hola infrastructure rather than isolated or incidental traffic.
Following these connections, affected customer environments showed downloads of suspicious executable files from rare external endpoints 188.241.219[.]55 and 184.241.218[.]111. Both endpoints have been flagged as potentially malicious by open-source intelligence (OSINT) [1][2].
These downloads were conducted using consistent user agents across impacted customers, specifically ‘Hola svc_js_win32/1.249.408’ and ‘Hola svc_js_win32/1.251.389’, suggesting a possible association with Hola-related activity.
Notably, this pattern aligns with recent reporting that, in some cases, Hola distributed an undeclared executable component, me[.]exe, which was later assessed to be a likely Monero-mining binary introduced via a compromised delivery pipeline [3].
Case Study 1
Darktrace first observed a new device on January 19, 2026, within a customer environment based in the Europe, Middle East, and Africa (EMEA) region. On the same day it appeared on the network, the device communicated with multiple pieces of Hola VPN-linked infrastructure before downloading a binary from a hola[.]org subdomain.
Figure 1: Cyber AI Analyst investigation highlighting Hola VPN service activity potentially associated with subsequent HTTP command-and-control (C2) connections.
Subsequent Darktrace telemetry revealed a recurring pattern of activity from the day the device was first observed through to March 4, 2026. During this period, the device repeatedly issued HTTP GET requests to the URI /bwfile?size=1048576, each returning a 200 OK response, indicating successful file retrieval.
This behavior was accompanied by a POST request to /bwfile, followed by an additional GET request for a significantly larger file at /bwfile?size=26214400, suggesting a deliberate and structured file transfer pattern.
Notably, the binary download activity was not tied to a single static host. Instead, it was observed across multiple URLs that changed over time while remaining within the same hola[.]org domain. This pattern suggests the use of rotating or distributed delivery infrastructure rather than a fixed endpoint.
Figure 2: Variation in URLs over time within the same hola[.]org domain, indicating the use of dynamically changing endpoints.
Across these events, the activity was consistently associated with the user agent Hola svc_js_win32/1.249.408, further linking the traffic to Hola-related service components. Amid these persistent and unusual connections, on February 22, Darktrace observed the device connecting to 188.241.219[.]55/proxy-peer-windows-amd64[.]exe, resulting in the download of an executable file.
Figure 3: File transfer event showing the download of an executable from the rare external endpoint 188.241.219[.]55.
Based on its file hash, the downloaded file was assessed as a likely Trojan downloader [4], with import hash (imphash) values showing similarities to samples linked to Vidar, Rhadamanthys, and Stealc according to OSINT [5]. Overall, this sequence of activity suggests that Hola-related connectivity may have been leveraged as part of a broader malware delivery chain.
Darktrace’s Autonomous Response
Due to the highly unusual activity observed, Darktrace Autonomous Response was triggered by the device’s behavior. However, as the customer deployment was configured in “Human Confirmation” mode, manual approval was required before any action could be taken.
Had the deployment been set to “Fully Autonomous” mode, Darktrace would have automatically:
Blocked connections to the associated ports and external endpoints
Prevented all outgoing network connections from the device
Enforced the device’s established ‘pattern of life’, allowing normal activity to continue while restricting any anomalous behavior
Figure 4: Example of a Darktrace Autonomous Response model highlighting the action that would have been taken, demonstrating how the system identifies anomalous behavior and applies targeted containment measures to restrict suspicious network activity.
Case Study 2
While the first case focused on anomalous activity from a newly observed device, Darktrace also identified cases in which devices had already been communicating with Hola-related endpoints prior to the suspected campaign. This may suggest pre-existing Hola usage within the environment, potentially increasing exposure and creating an avenue for subsequent suspicious activity.
One case involved three devices within a customer network based in the Americas (AMS). In this instance, a different payload was identified: me[.]exe, a potentially malicious cryptocurrency miner also referred to as HolaMonitorService[.]exe [6][7]. The downloads were observed from infrastructure similar to that seen in Case 1, including an IP address within the same 188.241.0.0/16 subnet.
Connections to *.hola[.]org, alongside the use of potential Hola-related user agents consistent with those in Case 1, were also identified, further suggesting a link between the observed activity and Hola-associated infrastructure.
Darktrace observed activity indicative of unusual VPN usage on the first affected device on February 2, followed by telemetry suggesting potential Tor usage. This was later followed by the download of me[.]exe on March 10 from 188.241.218[.]111. Notably, this device was the earliest among the three within the deployment to exhibit the presence of the suspicious executable.
Figure 5: Cyber AI Analyst detection highlighting the download of a suspicious executable from a similar external endpoint in a separate deployment.
On March 5, 2026, the second affected device exhibited a slightly different progression, initiating connections to http-test1[.]hola[.]org using the user agent ‘hola_get’. This activity was followed by the download of me[.]exe from the same endpoint on March 13, consistent with the broader pattern of Hola-related downloads observed across the environment.
Figure 6: Example of Hola VPN-related connectivity observed on the network prior to the suspected campaign, indicating pre-existing usage that may have contributed to subsequent activity.
The final affected device within this customer’s network demonstrated a more limited but related pattern, also downloading me[.]exe on March 17 using the same ‘hola_get’ user agent.
While the earlier Hola VPN usage observed across the deployment may not have been directly related to the suspected malware campaign, it may nonetheless have contributed to reduced visibility. The presence of pre-existing Hola-related traffic could have obscured malicious activity, making it more difficult to distinguish legitimate usage from attacker-driven behavior and, in turn, hindering the timely identification of the emerging compromise.
Darktrace’s Autonomous Response
For this deployment, the customer had their Autonomous Response capability configured in “Fully Autonomous” mode, allowing Darktrace to take action without human intervention. As a result, the system was able to autonomously disrupt the activity as soon as relevant events were identified through model detections.
Figure 7: Darktrace Autonomous Response actions taken against suspicious activity linked to Hola VPN.
Suspected cryptomining activity
As previously noted, some of the observed executable payloads appear to be linked to cryptomining malware. Across a subset of affected customer environments, this assessment was further supported by subsequent device activity consistent with Monero mining. Affected devices established follow-on connections to multiple external endpoints aligned with known mining infrastructure, indicating post-download execution.
Considering the broader sequence of activity, this pattern may point to a wider form of abuse in which legitimate VPN-related traffic is used to mask or facilitate malicious behavior following compromise.
On several devices, the download of executable files, including a newly observed peer[.]exe, was followed by alerts indicative of cryptocurrency mining activity. Mining-related credentials such as ‘x’ were observed using the Minergate protocol to communicate with endpoints within the 89.125.255.0/24 subnet and 188.241.218[.]111, the same endpoint involved in earlier download activity. Additional credentials appeared to reflect device-specific CPU identifiers, for example ‘12th Gen Intel(R) Core (TM) i5-1235U’.
Observed mining methods included login, submit, and job, consistent with active participation in a pool-based mining workflow rather than passive or incidental contact. The login method indicates that the host authenticated to the mining service as a worker, job reflects the assignment of computational tasks, and submit shows completed work being returned to the pool [8]. This sequence suggests that affected devices were actively contributing processing resources as part of an unauthorized distributed mining operation.
The presence of unauthorized cryptominers can lead to degraded system performance and reduced device stability. Beyond the immediate resource impact, such activity often serves as an indicator of a broader compromise rather than an isolated issue. This may increase the risk of further malware deployment, persistence mechanisms, and lateral movement, particularly in environments where the initial intrusion has not been fully contained.
Conclusion
Across affected environments, detections such as unusual VPN usage, connections to Hola infrastructure, anomalous HTTP activity, suspicious file downloads, and subsequent cryptomining behavior were linked into a single, evolving incident narrative. This aggregation provided a clearer view of attack progression, enabling security teams to understand not just isolated alerts, but the full sequence of compromise from initial contact through to post-exploitation.
Ultimately, these activities show that the risk posed by non-compliant software such as Hola VPN can extend far beyond simple policy violations. What began as traffic to Hola-related infrastructure was, in multiple cases, followed by behavior suggesting deliberate misuse, including suspicious executable downloads using Hola-related user agents and, in some instances, evidence of active cryptomining. These were not isolated anomalies, but elements of a broader pattern in which seemingly benign proxy or VPN-related communications may have created a pathway for malicious delivery and unauthorized resource exploitation.
The significance of this activity lies not only in the downloads or mining, but in what it reveals about an attacker’s ability to blend malicious operations into traffic associated with software that may already have a foothold in the environment. When unapproved software operates within an enterprise, it can reduce visibility, blur the distinction between legitimate and malicious traffic, and create opportunities to extend compromise in ways that are persistent and difficult to detect. Darktrace’s anomaly-based approach enables these behavioral distinctions to be identified, regardless of whether the device is new or long established within the network.
Credit to Min Kim (Associate Principal Analyst), Priya Thapa (Senior Cyber Analyst) Edited by Ryan Traill (Content Manager)
![Variation in URLs over time within the same hola[.]org domain, indicating the use of dynamically changing endpoints.](https://cdn.prod.website-files.com/626ff4d25aca2edf4325ff97/6a304bdeff111215c0436931_Screenshot%202026-06-15%20at%2012.00.43%E2%80%AFPM.png)
![File transfer event showing the download of an executable from the rare external endpoint 188.241.219[.]55.](https://cdn.prod.website-files.com/626ff4d25aca2edf4325ff97/6a304c2468d20aeff16a2721_Screenshot%202026-06-15%20at%2012.01.53%E2%80%AFPM.png)
