✨ No need to do from scratch. Deploy a Validated Containment Architecture built for your AI platform. →Deploy a Validated Containment Architecture for your AI platform. →A Validated Containment Architecture for your AI platform. →Validated Containment Architectures are here. →Contain Threats NowExplore✨
Financial Services
Breach intelligence, attack campaigns, and threat reports targeting the Financial Services sector.
Explore Other Sectors
Financial Services Threat Reports
Deepfake Voice Attacks: The Rising Threat in 2025
In March 2025, a finance director at a multinational firm in Singapore participated in a Zoom call with individuals appearing as her senior leadership team, including the CFO. Unbeknownst to her, all participants were AI-generated deepfakes. She authorized a $499,000 transfer before the fraud was detected. This incident mirrors a 2024 attack on Arup, where $25.6 million was stolen using similar deepfake techniques. The proliferation of deepfake technology has led to a 680% increase in voice deepfake incidents in 2025, with over 100,000 attacks recorded in the United States alone. The accessibility of these tools, which require minimal audio samples and no technical expertise, underscores the urgent need for organizations to implement robust verification protocols and employee training to mitigate such sophisticated social engineering threats.
3 months ago
Kill Chain
Anthropic's Claude Mythos AI Model Unveils Thousands of Zero-Day Vulnerabilities
In April 2026, Anthropic unveiled Claude Mythos Preview, an advanced AI model capable of autonomously identifying and exploiting zero-day vulnerabilities across major operating systems and web browsers. This model discovered thousands of high-severity vulnerabilities, including a 27-year-old bug in OpenBSD, and demonstrated the ability to chain multiple flaws to bypass security mechanisms. Due to the potential risks associated with its capabilities, Anthropic restricted access to Mythos, providing it only to select industry partners under Project Glasswing to allow for remediation before broader release. The emergence of AI models like Claude Mythos signifies a paradigm shift in cybersecurity, where the speed and scale of vulnerability discovery and exploitation are dramatically increased. This development underscores the urgent need for organizations to enhance their defensive strategies, prioritize rapid patch management, and adopt AI-driven security solutions to keep pace with evolving threats.
3 months ago
Kill Chain
Analyzing GitHub's March 2026 RCE Vulnerability (CVE-2026-3854)
In March 2026, GitHub identified a critical remote code execution (RCE) vulnerability (CVE-2026-3854) affecting its platforms, including GitHub.com and GitHub Enterprise Server. The flaw allowed users with push access to execute arbitrary commands on the server during a git push operation by exploiting unsanitized push options. GitHub promptly validated the issue, deployed a fix within two hours, and confirmed no evidence of exploitation. This incident underscores the importance of rigorous input sanitization and rapid response mechanisms in mitigating supply chain vulnerabilities. As software supply chains grow increasingly complex, organizations must prioritize proactive security measures to prevent similar threats.
3 months ago
Kill Chain
VECT 2.0 Ransomware: A New Threat to Data Integrity
In April 2026, the VECT 2.0 ransomware emerged, targeting Windows, Linux, and ESXi systems. Due to a critical flaw in its encryption implementation, files larger than 131KB are irreversibly destroyed, rendering recovery impossible even for the attackers. This flaw effectively transforms VECT 2.0 into a data wiper rather than traditional ransomware. ([gixtools.net](https://gixtools.net/feeds/items/vect-2-0-ransomware-irreversibly-destroys-files-over-131kb-on-windows-linux-esxi/?utm_source=openai)) The incident underscores the evolving nature of cyber threats, where flawed ransomware can lead to permanent data loss. Organizations must prioritize robust backup strategies and incident response plans to mitigate such risks.
3 months ago
Kill Chain
After Mythos: New Playbooks For a Zero-Window Era
In April 2026, Anthropic's AI model, Claude Mythos, autonomously identified thousands of zero-day vulnerabilities across major operating systems and web browsers. This unprecedented capability led to the formation of Project Glasswing, a collaborative initiative involving tech giants like Apple, Google, and JPMorgan Chase, aiming to patch vulnerabilities faster than AI can discover them. However, unauthorized access to Mythos raised significant security concerns, highlighting the potential risks of such powerful AI tools. ([anthropic.com](https://www.anthropic.com/glasswing?utm_source=openai)) The incident underscores the urgent need for robust security protocols in AI development and deployment. As AI models become more sophisticated, ensuring their secure use is paramount to prevent potential misuse and safeguard critical infrastructure.
3 months ago
Kill Chain
Microsoft Entra ID Agent ID Administrator Role Privilege Escalation Vulnerability
In March 2026, a critical vulnerability was identified in Microsoft Entra ID's Agent ID Administrator role, designed to manage AI agent identities. This flaw allowed users with this role to take over arbitrary service principals by assigning themselves as owners and adding new credentials, potentially escalating privileges to the Global Administrator level. Microsoft addressed the issue by April 9, 2026, restricting the role's permissions to prevent such unauthorized access. This incident underscores the importance of stringent role scoping and continuous monitoring of privileged accounts to prevent similar security breaches in the future.
3 months ago
Kill Chain
UNC6692's 'Snow' Malware: A New Era of Social Engineering Attacks
In December 2025, the threat actor UNC6692 initiated a sophisticated attack combining social engineering, cloud service exploitation, and custom malware. The campaign began with a flood of spam emails to targets, followed by impersonation of IT helpdesk staff via Microsoft Teams. Victims were deceived into downloading a fake 'Mailbox Repair Utility,' which installed the 'Snow' malware suite, including Snowbelt, Snowglaze, and Snowbasin. This malware facilitated unauthorized access, credential theft, and lateral movement within enterprise networks. The attackers leveraged AWS S3 buckets for payload delivery and command-and-control infrastructure, effectively bypassing traditional security measures. ([darkreading.com](https://www.darkreading.com/cloud-security/unc6692-social-engineering-malware-cloud-abuse/?utm_source=openai)) This incident underscores the evolving tactics of cyber adversaries who exploit trusted cloud services and communication platforms to infiltrate organizations. The use of legitimate cloud infrastructure for malicious purposes highlights the need for enhanced monitoring of cloud-based activities and user education to recognize sophisticated social engineering attempts.
3 months ago
Kill Chain
GlassWorm Campaign Escalates with Malicious VS Code Extensions
In April 2026, the GlassWorm campaign escalated by deploying 73 malicious Visual Studio (VS) Code extensions on the Open VSX marketplace. These extensions, initially appearing benign, were later updated to deliver self-replicating malware, compromising developer environments and potentially poisoning the software supply chain. The malware utilized techniques such as external payload retrieval and bundled native binaries, acting as thin loaders to evade detection. This approach allowed attackers to access sensitive information, including source code, credentials, and internal systems, posing significant risks to organizations relying on these tools. The resurgence of GlassWorm highlights the evolving nature of supply chain attacks, emphasizing the need for continuous monitoring of software dependencies. Organizations must implement stringent security measures, such as verifying the authenticity of extensions, auditing installed tools for recent updates, and educating developers on the risks associated with third-party software. This incident underscores the critical importance of securing the software development lifecycle to prevent widespread compromise.
3 months ago
Kill Chain
Understanding the 2026 AWS Cognito Refresh Token Abuse Incident
In March 2026, AWS updated its Threat Technique Catalog to highlight a significant security concern: the abuse of Amazon Cognito refresh tokens. Threat actors have been exploiting long-lived refresh tokens to maintain unauthorized access to AWS environments. By obtaining a valid refresh token—through methods like credential theft or compromised client-side storage—attackers can continuously generate new access and ID tokens without re-authentication, effectively establishing a persistent foothold in the system. This technique allows them to operate undetected, as the legitimate user's session remains unaffected. The default lifespan of these tokens is 30 days, but they can be configured for up to 10 years, amplifying the potential risk. ([aws-samples.github.io](https://aws-samples.github.io/threat-technique-catalog-for-aws/Techniques/T1098.A006.html?utm_source=openai)) This incident underscores the evolving tactics of cyber adversaries who leverage legitimate cloud service functionalities to evade detection. Organizations must reassess their security postures, particularly concerning token management and monitoring, to mitigate such stealthy persistence mechanisms.
3 months ago
Kill Chain
Security Breach: Unauthorized Access to Anthropic's Claude Mythos AI Model
In April 2026, Anthropic's advanced AI model, Claude Mythos, designed for cybersecurity applications, was accessed without authorization through a third-party vendor environment. The breach occurred on the same day the model was announced, with individuals from an online forum exploiting the access. Anthropic is investigating the incident and has not found evidence of broader system compromise. This incident underscores the challenges in securing powerful AI models, especially when third-party vendors are involved. It highlights the need for stringent access controls and monitoring to prevent unauthorized access to sensitive technologies.
3 months ago
Kill Chain
Critical Remote Code Execution Vulnerability in GitHub Enterprise Server (CVE-2026-3854)
In March 2026, a critical vulnerability (CVE-2026-3854) was identified in GitHub Enterprise Server, allowing authenticated users with push access to execute arbitrary code on the server through a crafted 'git push' command. The flaw stemmed from improper sanitization of user-supplied push option values, which were incorporated into internal service headers without adequate validation. This oversight enabled attackers to inject malicious metadata fields, leading to remote code execution. GitHub promptly addressed the issue by releasing patches for affected versions, including 3.14.25, 3.15.20, 3.16.16, 3.17.13, 3.18.7, and 3.19.4. ([nvd.nist.gov](https://nvd.nist.gov/vuln/detail/CVE-2026-3854?utm_source=openai)) This incident underscores the critical importance of rigorous input validation and prompt patch management in safeguarding software supply chains. As organizations increasingly rely on platforms like GitHub for code collaboration and deployment, ensuring the security of these infrastructures is paramount to prevent potential exploitation and maintain trust in software development processes.
3 months ago
Kill Chain
LofyGang's LofyStealer: A New Threat to Minecraft Players in 2026
In April 2026, the Brazilian cybercrime group LofyGang re-emerged after a three-year hiatus, launching a campaign targeting Minecraft players with a new malware known as LofyStealer. Disguised as a Minecraft hack named 'Slinky,' the malware uses the official game icon to deceive users into execution. Once activated, it deploys a JavaScript loader that installs LofyStealer ('chromelevator.exe') directly into the system memory. This stealer harvests sensitive data—including cookies, passwords, tokens, credit card information, and International Bank Account Numbers (IBANs)—from various web browsers such as Google Chrome, Microsoft Edge, Brave, Opera, Mozilla Firefox, and Avast Browser. The exfiltrated data is then transmitted to a command-and-control server controlled by the attackers. This incident underscores a significant shift in LofyGang's tactics from previous methods like typosquatting on npm packages to a malware-as-a-service (MaaS) model, offering both free and premium tiers. The campaign highlights the persistent threat posed by cybercriminals exploiting trusted platforms and popular games to distribute malicious software, emphasizing the need for heightened vigilance among users and robust security measures to protect sensitive information.
3 months ago
Kill Chain
Stop Active Cloud Data Exfiltration
Aviatrix Breach Lock helps teams instantly identify what data is leaving the environment, from which workload, and where it’s going — during an active breach.
Looking for threats in a different sector?
Browse All Threat Reports