karlo.zanki@reversinglabs.com (Karlo@Blog (Main)
//
Cybersecurity experts are raising alarms over the increasing use of artificial intelligence for malicious purposes. ReversingLabs (RL) researchers recently discovered a new malicious campaign targeting the Python Package Index (PyPI) that exploits the Pickle file format. This attack involves threat actors distributing malicious ML models disguised as a "Python SDK for interacting with Aliyun AI Labs services," preying on users of Alibaba AI labs. Once installed, the package delivers an infostealer payload hidden inside a PyTorch model, exfiltrating sensitive information such as machine details and contents of the .gitconfig file. This discovery highlights the growing trend of attackers leveraging AI and machine learning to compromise software supply chains.
Another significant security concern is the rise of ransomware attacks employing social engineering tactics. The 3AM ransomware group has been observed impersonating IT support personnel to trick employees into granting them remote access to company networks. Attackers flood an employee's inbox with unsolicited emails and then call, pretending to be from the organization's IT support, using spoofed phone numbers to add credibility. They then convince the employee to run Microsoft Quick Assist, granting them remote access to "fix" the email issue, allowing them to deploy malicious payloads, create new user accounts with admin privileges, and exfiltrate large amounts of data. This highlights the need for comprehensive employee training to recognize and defend against social engineering attacks. The US Department of Justice has announced charges against 16 Russian nationals allegedly tied to the DanaBot malware operation, which has infected at least 300,000 machines worldwide. The indictment describes how DanaBot was used in both for-profit criminal hacking and espionage against military, government, and NGO targets. This case illustrates the blurred lines between cybercrime and state-sponsored cyberwarfare, with a single malware operation enabling various malicious activities, including ransomware attacks, cyberattacks in Ukraine, and spying. The Defense Criminal Investigative Service (DCIS) has seized DanaBot infrastructure globally, underscoring the severity and scope of the threat posed by this operation. References :
Classification:
@cyberpress.org
//
A new variant of the Lumma Stealer malware has been identified, showing significant advancements in its stealth and persistence. Researchers at the Trellix Advanced Research Center analyzed the new variant, discovering features such as code flow obfuscation and dynamic API resolution that help it evade detection. Lumma Stealer, originally introduced in 2022, has rapidly evolved and poses a serious threat to personal and organizational data by targeting sensitive information stored on infected systems.
Lumma Stealer, also known as LummaC2, has gained popularity in underground forums with over a thousand active subscribers as of March 2025. The malware uses deceptive methods such as fake CAPTCHA pages, mimicking Google reCAPTCHA or Cloudflare challenges, to trick users into executing malicious commands. These fraudulent pages are often hosted on compromised websites offering pirated content or cryptocurrency services, enticing unsuspecting users to initiate the infection chain. The malware's infection chain is complex and difficult to detect. It involves downloading a .zip file, extracting the malware, and establishing persistence through the Windows Registry's Run key. More advanced attacks hide the malware within seemingly harmless .mp3 or .png files, triggered via the mshta.exe HTML application engine, deploying layers of encryption, anti-debugging techniques, and detection evasion mechanisms. The stealer targets sensitive data, including cryptocurrency wallet credentials, 2FA codes, browser-stored passwords, and financial information, which it transmits to attacker-controlled domains. References :
Classification:
Andres Ramos@Arctic Wolf
//
A resurgence of a fake CAPTCHA malware campaign has been observed, with threat actors compromising widely used websites across various industries. They are embedding a fake CAPTCHA challenge that redirects victims to a site triggering PowerShell code execution. This campaign exploits social engineering tactics and fake software downloads to deceive users into executing malicious scripts.
This tactic is also utilized with fake captchas which resemble legitimate sites. When users attempt to pass the captcha, they are prompted to execute code that has been copied to their clipboard. The OBSCURE#BAT malware campaign is a major cybersecurity threat to both individuals and organizations, primarily due to its ability to compromise sensitive data through advanced evasion techniques, including API hooking. This allows the malware to hide files and registry entries, making detection difficult. References :
Classification: |