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
5 Things You Must Know Now About the Coming EU AI Regulation
FunkSec: An AI-Centric and Affiliate-Powered Ransomware Group
Image Sources:
Back to Blog Posts
Podcast
Cyber Campfire: June Threat Trends & Insights
In this episode of Arete’s Cyber Campfire podcast, our Threat Intelligence Team discusses trends, statistics, and emerging threat actors from June 2026. Tune in for firsthand insights on today’s threat landscape that can enhance your approach to cyber risk.
Article
Current: Misconfigured Server Reveals Multiple Phishing Campaigns
Researchers uncovered three Microsoft 365 phishing campaigns after discovering a misconfigured HTTP server that exposed a threat actor’s phishing tools, configurations, victim logs, and operational infrastructure. The investigation revealed two operators using custom Evilginx-based adversary-in-the-middle (AiTM) phishing frameworks to steal authentication sessions and bypass MFA protections. A third operator abused Microsoft’s OAuth Device Code flow by tricking victims into completing authentication through Microsoft’s legitimate login process, allowing attackers to obtain valid access and refresh tokens without stealing passwords. The largest campaign compromised at least 218 victims across 12 countries, primarily targeting corporate Microsoft 365 accounts, with refresh tokens enabling persistent access. Researchers also identified the use of remote management tools for maintaining access, AI-assisted development of phishing tooling, and links to broader Phishing-as-a-Service (PhaaS) ecosystems, highlighting how commodity frameworks and automation are lowering the barrier for sophisticated identity-based attacks.
What’s Notable and Unique
The campaigns demonstrate the continued shift from traditional credential theft to identity- and token-based compromise, in which attackers target authentication sessions, cookies, and OAuth tokens rather than relying solely on password theft. By obtaining valid authentication artifacts, attackers can access Microsoft 365 environments while bypassing conventional MFA protections.
The use of Evilginx-based AiTM frameworks highlights how publicly available phishing tools have evolved into effective MFA-bypass platforms. These frameworks proxy legitimate login pages, capture authenticated sessions, and allow attackers to reuse valid sessions without requiring victims’ credentials again.
The OAuth Device Code flow abuse represents a different attack method, in which attackers leverage a legitimate Microsoft authentication mechanism to obtain tokens after victims approve the request. This approach bypasses password theft entirely by abusing trusted authentication workflows.
The exposed infrastructure provided insight into attacker operations, including phishing configurations, victim tracking data, remote administration tools, and evidence of AI-assisted development. The findings demonstrate how phishing ecosystems now provide reusable tooling and automation that reduce the technical expertise required to conduct large-scale identity attacks.
Analyst Comments
These Microsoft 365 phishing campaigns demonstrate the ongoing evolution of identity-focused attacks, as adversaries increasingly target authentication sessions, tokens, and legitimate cloud authorization mechanisms to bypass traditional security controls. The combination of AiTM phishing, OAuth abuse, remote management tools, and PhaaS infrastructure enables threat actors to maintain persistent access while blending into legitimate cloud activity.
The growing availability of phishing frameworks and AI-assisted attack development highlights the need for continuous identity security improvements beyond traditional phishing defenses. Arete recommends that organizations strengthen identity monitoring, adopt phishing-resistant MFA where possible, review OAuth application permissions, monitor token activity, and implement conditional access policies to detect and restrict suspicious Microsoft 365 authentication behavior. Security teams should also monitor post-compromise indicators such as abnormal mailbox access, suspicious application consent, unauthorized remote tools, and unusual sign-in activity.
Sources
One Misconfigured Server, Three Active Campaigns: Full exposure of three AiTM Phishing Operators
Misconfigured Server Reveals Three Evilginx Phishing Operations Targeting Microsoft 365
Podcast
Digital Forensics Meets the Courtroom: Insights on Expert Witness and Cyber-Related Litigation
In this episode of Bytes of Insight, host Vinny Sakore and special guest Dr. Bruce Hartley, Managing Director and Co-Founder of Advisory Services at Arete, discuss the evolving role of digital forensics and expert witness work in cybersecurity litigation. Tune in for firsthand insights on how Arete conducts post-breach investigations, what it takes to undergo cross-examination on the stand, and how emerging threats like AI-driven social engineering and third-party breaches are reshaping the advisory landscape.
Article
Helix Extortion Group Debuts in SharePoint Data Theft Attacks
A newly identified data-extortion group, Helix, is reportedly conducting targeted campaigns by exploiting identity-based attack techniques to gain access to Microsoft 365 environments and exfiltrate data from SharePoint Online. The group leverages voice phishing (vishing), device code phishing, and multi-factor authentication (MFA) abuse to compromise user accounts and establish unauthorized access.
Helix threat actors reportedly initiate intrusions through vishing campaigns, impersonating trusted individuals—often the target's manager—and leveraging caller ID spoofing to enhance credibility.
The actors then attempt to persuade victims to complete device code phishing workflows, enabling unauthorized access to corporate accounts and Microsoft 365 resources.
Following successful authentication, the actors conduct reconnaissance and enumeration of SharePoint environments before identifying and exfiltrating sensitive data for extortion purposes.
Helix's variable dwell times suggest a flexible, target-specific operational approach, enabling the group to adapt its intrusion timeline to victim environments to maximize data theft and extortion opportunities while minimizing the likelihood of detection. Helix's operations also appear to be focused on large-scale data theft and extortion, with an emphasis on automated SharePoint data exfiltration following successful credential or session compromise.
Overlaps with Other Extortion Groups
Researchers assess that Helix likely emerged from or has operational overlap with the ShinyHunters and BlackFile/Redact cybercriminal ecosystems. ShinyHunters historically focuses on compromising websites, developer repositories, and exposed credentials or API keys to gain access to corporate cloud environments, monetizing stolen victim data on the dark web. BlackFile utilizes identity-centric and social engineering–driven methods of intrusion, relying less on traditional malware and instead leveraging credential theft, privilege escalation, rapid data exfiltration, and coercive tactics such as swatting to maximize extortion pressure. In May 2026, BlackFile announced that it would be rebranding as “Redact” (which may have later rebranded again to “Pink”), yet despite the rebrands, threat actors operating within these groups continued to exhibit similar operational models, victimology, and extortion methodologies.
Analyst Comments
Collectively, the similarities between the groups suggest the possibility of a shared criminal ecosystem, affiliate network, or loosely connected threat cluster rather than entirely independent operations. The emergence of Helix further reflects the broader trend of extortion groups rebranding or fragmenting while retaining proven attack techniques and operational infrastructure to support their criminal activities. Despite suspicion of these groups’ connectivity, the exact nature of the relationship remains unconfirmed. As Helix focuses on identity-based attack techniques to gain access to Microsoft 365 environments, Arete recommends that organizations remain vigilant in using MFA, enforcing strong identity security controls, continuously monitoring authentication activity, and enhancing user awareness of social engineering tactics.
Sources
Helix, a New Name in the Data Extortion Ecosystem?
BlackFile actively extorting data-theft victims in retail and hospitality sector
"Pink" Data Extortion Group Hunting With Evasive Phishing Kits



