EXPLORE

Article

AI Deep Dive Part 3: Understanding Biases & How Threat Actors Use AI

Arete Analysis

Cybersecurity 101

In the final installment of Arete’s Artificial Intelligence (AI) deep dive, we will explore different types of biases that can be present in AI and how threat actors are able to leverage AI in cybercrime. Inherent biases are present in both human thought processes and AI models, which often influence the data used for training and the algorithms themselves. Additionally, while many AI tools are designed and intended for legitimate purposes, users should be aware that they are also continuously used by cybercriminals to enable operations.

Biases in AI

Humans all have inherent biases in our thought processes, which is also true in AI models. The data used to train models, and the models themselves are victims of biases which influence the final AI product and its responses. When considering biases within AI, algorithmic biases are top of mind. Algorithmic bias outlines the systematic and repeatable errors in a computer system that create unfair outcomes. A study conducted by the European Union’s Artificial Intelligence Act (EU AIA) did a great job of outlining the real-world implications of biases infiltrating AI.

Source: 5 Things You Must Know Now About the Coming EU AI Regulation

What begins with minimal risk at the bottom of this graphic, with biases influencing video gameplay and spam filters, quickly progresses to high-impact areas like transportation systems, justice systems, and even facial recognition. A real-world example of this would be the situation where an individual is wrongly accused of a crime due to biases in AI based on their age, gender, or skin color, evidence that algorithmic biases need to be understood and mitigated to correct incorrect outcomes facilitated by potential false information generated due to algorithmic bias.

AI in the Wrong Hands

Just as AI enables the work of professionals across many industries, cybercriminals have also begun to exploit this technology. While most public AI models have filters in place to prevent the malicious use of the models, these filters can often be bypassed to create convincing malicious social engineering content. Additionally, some threat actors, like the ransomware group Funksec, have developed their own AI models without these limitations.

During an interview, an excerpt of which is shown below, the leader of Funksec emphasized their ability to act as developers and bring high-level ideas while AI acts as the programmer, enabling their ideas using their proprietary WormGPT module. This AI use case means less technical cybercriminals are becoming increasingly able to write malware without the scripting knowledge typically required.

An excerpt from an interview with a Funksec operator. Source: Threat Actor Interview: Spotlighting on Funksec Ransomware Group

Here’s an example of a WormGPT response when asked for a webshell and C++ code.

Source: Threat Actor Interview: Spotlighting on Funksec Ransomware Group

Conclusion

In the right hands, AI is a powerful tool for productivity and advancement, but the potential risks associated with the growth of AI in connection with illegal activity such as cybercrime should be carefully monitored and addressed. Both the unintended negative effects of biases and the use of AI by cybercriminals highlight the challenges for evolving safeguards and controls protecting correct and ethical AI use.

Sources

Image Sources:

Back to Blog Posts

Article

Top 9 Security Controls for SMBs

In today’s complex cyber threat landscape, small and medium-sized businesses (SMBs) should implement a layered, multi-pronged approach to security. Here are some proven, effective controls to consider:

1.        Implement MFA for All Users
Multi-factor authentication (MFA) is an effective and accessible way to protect sensitive data, comply with regulatory standards, and prevent unauthorized access to critical systems. By ensuring that users not only use a password but also a secure token, an organization can significantly reduce phishing attempts, credential stuffing attacks, and ransomware incidents. MFA should be enabled for email, VPN, and critical system access.

2.        Update Patch Management Programs
A modern patching program should include policies and mechanisms to manage software updates in a timely manner. Patching efforts should address not just operating system updates, but also commonly utilized software within the environment. An active patching plan aims to reduce the mean time to patch and provide metrics on existing patch efforts. In the event of a high-severity vulnerability, a mature patching program should also be able to identify the risks and exposure for a given set of systems based on the average time to patch. This helps stakeholders make informed decisions when a new exploit is released.

3.        Leverage a Properly Managed Endpoint Detection and Response (EDR) Tool
With real-time threat detection and advanced visibility, endpoint detection and response (EDR) solutions can enhance an organization’s ability to detect and investigate potential malicious activity, including ransomware, lateral movement, fileless attacks, and more. Effective management of EDR should include proper configuration, regular agent updates, and established procedures for responding to alerts.

4.        Enhance EDR with Frontline Incident Intelligence
While EDR solutions are an important component of a security program, threat actors are continually adapting their techniques to evade detection, and as a result, standard detection methods may not always identify emerging threats or novel attack behaviors. To strengthen protection, organizations can supplement EDR tools with custom detection rules informed by incident response data. These rules can help identify indicators of compromise observed during cyber incidents, providing an additional layer of visibility beyond traditional detections.

Arete Bloktd℠ illustrates this approach by transforming threat intelligence and malware analysis into curated detection rules for EDR tools. Applying frontline intelligence to security tools can result in earlier detection, less investigative burden, and a more confident response. 

5.        Create and Continuously Test an Incident Response (IR) Plan
An incident response plan can help an organization effectively identify, respond to, and recover from a cybersecurity incident while minimizing disruption and downtime. An IR plan defines clear roles, escalation paths, communication strategies, and decision frameworks for both internal and external teams. To be effective, these plans should be regularly tested with post-incident reviews or tabletop exercises involving key stakeholders.

6.        Maintain Encrypted and Tested Data Backups
In the case of a cyberattack, backups can help organizations maintain business continuity, support ransom negotiations with threat actors, and contribute to a timely and successful recovery. Encrypting backups makes it more difficult for threat actors to access or alter them, and regular testing ensures they remain accessible during a security incident.

7.        Conduct Employee Security Awareness Training
Human error is a leading cause of cybersecurity incidents, and an organization’s employees are its first line of defense against cyber threats. Security awareness training reinforces the importance of data confidentiality, supports regulatory compliance, and helps employees recognize and understand red flags such as suspicious emails, malicious links, and social engineering attempts.

8.        Assess and Address Existing Risk and Vulnerabilities
To maintain a proactive approach to risk management and improve resilience, organizations should regularly assess existing vulnerabilities through expert-led Vulnerability Assessments, Penetration Testing, and Threat Hunting Assessments. These exercises can help identify security gaps, support regulatory compliance, and prevent potential incidents. During remediation, prioritize high-impact vulnerabilities and make a plan for periodic assessments.

9.        Apply Least-Privilege Access Controls
Employees should have access only to the systems, applications, and data necessary to perform their jobs. By removing unnecessary administrative privileges and regularly reviewing user permissions, organizations can reduce the risk of unauthorized access and limit the damage caused by compromised accounts or insider threats.

Organizations should regularly review their cybersecurity programs, stay informed about emerging threats, and evaluate whether their security controls are adapting to the evolving threat landscape.

Article

Ransomware Trends & Data Insights: August 2026

Although INC Ransom was the most active ransomware threat group observed in August 2026, overall activity remained relatively distributed across the threat landscape, with approximately 30 unique ransomware and extortion groups, including both established and newly emerged actors, observed during the month. Alongside INC Ransom, Storm, DragonForce, Qilin, and The Gentlemen were among the top five most active threat groups, underscoring the continued diversity and competitiveness of the ransomware ecosystem.

Figure 1. Activity from the top 5 threat groups in August 2026 

Throughout the month, analysts at Arete identified several trends behind the threat actors perpetrating cybercrime activities: 

  • August 2026 was a significant month for the emergence of new threat actors, with Arete identifying several previously untracked ransomware groups, including Storm, Dark Project, DireWolf, Morpheus, Orova, Spirals, Wallstreet Team, and Zawoo Team. This trend underscores the ongoing evolution of the threat landscape and the low barriers to entry for new cybercriminal operations. 

  • Arete observed a continued use of AI-enabled capabilities across the ransomware ecosystem during August. Similar to Fulcrumsec and Setra, the newly emerged Zawoo Team appears to leverage AI-assisted analysis of stolen data, publishing highly structured and detailed victim assessments on its data leak site (DLS). Additionally, both the Aurora and Storm ransomware groups have been observed utilizing AI-driven or AI-assisted chat representatives during victim communications. These developments suggest that threat actors are increasingly incorporating AI to enhance data analysis, automate victim engagement, and improve the effectiveness and scalability of extortion operations. 

  • In August, Arete also observed that despite sustained law enforcement disruption efforts, LockBit continues to demonstrate operational resilience, likely through the expansion of its affiliate ecosystem and the evolution of its tactics. Recent activity indicates an increased reliance on data theft and exfiltration-only operations, with observed victimology suggesting a notable focus on organizations within the financial sector. 

Sources

  • Arete Internal

Report

Arete's H1 2026 Crimeware Report

Explore intelligence and operational data from the frontlines of ransomware and extortion response as we detail the most significant trends and developments in the cyber threat landscape in the first half (H1) of 2026.

Article

WinRAR Path-Traversal Vulnerability Exploited in Ransomware Attacks

CISA has warned that CVE-2025-8088, a path-traversal vulnerability in WinRAR, is being exploited in ransomware attacks. The vulnerability allows attackers to use specially crafted archive files to place malicious executables in locations such as the Windows Startup folder, enabling code execution when the system restarts. WinRAR addressed this vulnerability in version 7.13, but exploitation has continued due to challenges in updating the software. CISA's updated KEV entry now specifically identifies the vulnerability as being used in ransomware attacks. 

What’s Notable and Unique

  • The vulnerability was initially exploited as a zero-day by the Russia-linked RomCom group, enabling malware execution at user logon and supporting credential theft, cyberespionage, and ransomware-related activities.

  • CVE-2025-8088 continues to be actively exploited by both government-backed and financially motivated threat actors. Observed campaigns have been attributed to Russian- and Chinese-linked groups targeting several organizations.

  • Phishing campaigns leveraging weaponized RAR attachments exploited CVE-2025-8088 to silently place malicious payloads in autorun locations using directory traversal and Alternate Data Streams (ADS). 

  • The widespread exploitation of CVE-2025-8088 highlights strong demand for effective WinRAR exploits, with multiple underground vendors previously supplying exploit capabilities to a diverse range of threat actors. One such vendor, zeroplayer, advertised a WinRAR exploit in July 2025. 

Analyst Comments

Organizations should immediately upgrade WinRAR to version 7.13 or later and conduct an enterprise-wide inventory to identify unmanaged installations. Given that WinRAR lacks automatic updates, Group Policy support, and integration with common enterprise patch management platforms, manual verification and remediation may be required to ensure all vulnerable instances are addressed. Security teams should also enhance defenses against archive-based phishing attacks by blocking or scrutinizing unsolicited compressed files, monitoring for suspicious file creation in Windows Startup directories, and detecting abuse of ADS. With CISA now linking CVE-2025-8088 to ransomware activity, organizations should treat remediation as a high priority and actively monitor for indicators of WinRAR-based exploitation across their environments.

Sources

  • CISA Adds Three Known Exploited Vulnerabilities to Catalog

  • WinRAR vulnerability being exploited in ransomware attacks, CISA warns