Search This Blog

Powered by Blogger.

Blog Archive

Labels

Footer About

Footer About

Labels

Showing posts with label Authentication Tokens. Show all posts

New CSS Attacks Expose Webmail Users to Passord and Token Theft


In new research, CSS-based attacks have been discovered that can bypass security protections in webmail services, allowing attackers to steal passwords, authentication tokens, and other sensitive data. In order to demonstrate the ability of malicious email content to interact with trusted elements within a webmail interface, these techniques demonstrate how their contents can escape their intended boundaries. 

Gareth Heyes, PortSwigger researcher, presented the study at Black Hat USA 2026. The research examined attack chains with Outlook, Gmail, Fastmail, Proton Mail, Yahoo Mail, and AOL Mail. There has been no report of proof-of-concept attacks being used in the wild, however researchers have warned that vulnerabilities in the way webmail platforms handle HTML and CSS may pose serious security risks. 

In one of the most significant demonstrations, researchers were able to disguise a select element as a password field by combining HTML elements, CSS manipulation, and browser behavior. Through the use of this technique, a victim could enter a password into what appeared to be a genuine Microsoft login interface and be captured as a result. 

An underlying problem stems from the fact that webmail platforms allow untrusted HTML and CSS to be displayed within an otherwise trusted interface. It was discovered that attackers could either take advantage of CSS features already allowed by webmail services or exploit the gap between what the sanitizer approves and what the browser actually renders by exploiting the sanitizer. There are also instances in which applications can utilize JavaScript to create new DOM elements with CSS properties that weren't included in the original allowlist of the sanitizer, resulting in the malicious email content escaping its intended boundaries.

Researchers discovered that Yahoo Mail and AOL Mail were vulnerable to a different vulnerability involving pasted HTML Researchers showed how CSS remains active for a short period of time before sanitization allows the attacker to extract portions of an email login token. A 12-character token can then be reconstructed and used to access the victim's account. A CSS selector can be used to determine which digits appear in an email when Content Security Policy prevents an email from making conventional external requests. 

Using CSS selectors, an attacker can identify which digits appear in an email as well as how frequently they occur. By carefully positioning the links so that only the matching option is visible, the attacker-controlled server will receive the inferred information with only one click, without the need for JavaScript. During the Gmail attack, users were required to ask Cowork to process their affected emails, after which the malicious instructions embedded within the message influenced the way the AI system handled the account information. 

During the demonstration, prompt injection, along with email content, demonstrated how the legitimate access of an AI assistant can be turned into a path to expose sensitive tokens, particularly when the assistant has the ability to read messages and write email drafts. A similar attack against the Atlas browser of OpenAI was demonstrated with Fastmail.

In CSS techniques, malicious instructions are concealed from the user while being visible to the artificial intelligence system, demonstrating that differences in how web content is interpreted by human and machine are potentially dangerous, leading to new attack opportunities. Moreover, the researchers identified techniques for manipulating trusted interface actions, bypassing certain content restrictions, and revealing information, such as the time an email was viewed or the IP address of the recipient, in one Proton Mail demonstration. 

Researchers also demonstrated the potential for revealing the recipient's IP address using Proton Mail's tracker-protection mechanisms. As a result of Proton Mail's tracker-protection mechanism, email senders are not able to obtain the IP address of a user and the precise time of email open, the demonstrated bypass illustrates yet again the vulnerability of CSS and webmail rendering behavior to undermine privacy. 

Fastmail was reported to have fixed two CSS mutation vulnerabilities at the time of publication, while the Proton Mail proxy bypass demonstrated at the time was not observed during retesting. However, the Outlook label-jacking technique and Gmail's image-set() bypass remain effective as of August 6. Additionally, the research did not establish whether all aspects of the Outlook password capture chain had been resolved. 

According to the researchers, HTML email should be contained within sandboxed iframes, strict CSS allowlists should be applied, dangerous selectors and select menus should be blocked, custom attributes should be examined for CSS-based attack gadgets and attacker-controlled image requests should be prevented. As traditional content-sanitization defenses face increasingly sophisticated CSS-based attacks, webmail security is becoming increasingly complex. 

With the advent of artificial intelligence assistants having access to email inboxes and other connected services, providers should strengthen their isolation and rendering controls in order to prevent malicious email content from becoming an avenue for credential theft as well as data exposure.


Researcher Reveals VS Code Flaw That Could Expose GitHub Access Tokens Through a Single Click

A publicly disclosed security flaw affecting the browser-based version of Visual Studio Code has drawn attention from developers after a researcher demonstrated how attackers could potentially obtain GitHub authentication tokens through a single user interaction.

The issue was disclosed by security researcher Ammar Askar, who published technical details alongside proof-of-concept code showing how the vulnerability could be abused. At the time of disclosure, no CVE identifier had been assigned and Microsoft had not released an official software patch.

According to Askar's analysis, the weakness exists within github.dev, GitHub's web-based development environment that allows users to work with repositories directly from a browser using technology derived from Visual Studio Code. The attack takes advantage of the way VS Code's webview components communicate with the main editor environment.

Webviews are embedded browser windows used by extensions and web applications to display interactive content. While these components are designed to operate within restricted environments, the researcher found a method to abuse the message-passing mechanism that connects a webview to the editor interface.

The published demonstration shows how malicious JavaScript running inside a webview can trigger actions within the main editor window. By simulating keyboard input and user activity, the code can install a malicious extension without requiring the victim to manually perform the installation process.

Once deployed, the extension is capable of extracting a GitHub OAuth token that is transmitted when users access github.dev. OAuth tokens act as authorization credentials that allow applications to interact with GitHub services on behalf of authenticated users.

According to the researcher, the security concern extends beyond access to a single repository. The token passed to github.dev can inherit the permissions associated with the user's GitHub account, potentially granting access to every repository available to that account, including private projects.

Using the proof-of-concept attack, a malicious extension can retrieve the token and communicate with GitHub's API. This allows an attacker to identify repositories accessible to the compromised account and gather information about private development resources.

Askar argued that the broad permissions associated with the token significantly increase the potential impact of exploitation because access is not limited to the repository that initially triggered the github.dev session.

To reduce exposure while no official fix was available, the researcher advised users to clear cookies and locally stored site data associated with github.dev. Removing this stored data forces additional authentication checks that can help expose suspicious sign-in attempts.

After clearing the stored information, users attempting to access github.dev through a malicious link would be more likely to encounter a warning indicating that the GitHub Repositories extension is requesting authorization through GitHub. Such prompts can serve as an indication that unexpected account access is being requested.

The disclosure also highlighted ongoing tensions surrounding vulnerability reporting processes. Askar stated that GitHub was notified approximately one hour before publication of the research. He described the disclosure as a deliberate decision to release the information publicly rather than pursue a lengthy coordinated disclosure process.

The researcher cited previous interactions involving another VS Code vulnerability that he reported through Microsoft's security channels. According to his account, the issue was later addressed without attribution and was classified as having no security impact despite his concerns regarding its implications.

Askar said that experience influenced his decision to publicly disclose future VS Code security findings rather than continue working through Microsoft's reporting process.

The incident follows several other public disclosures involving Microsoft products by an independent researcher operating under the online alias "Nightmare Eclipse." Over recent months, that researcher has released details regarding multiple unpatched vulnerabilities affecting Windows and related Microsoft technologies, including flaws known as BlueHammer, RedSun, GreenPlasma, MiniPlasma, YellowKey, and UnDefend.

Some of those vulnerabilities were later reported as being actively exploited, further intensifying discussions within the security community about vulnerability handling, disclosure timelines, and communication between vendors and independent researchers.

Microsoft previously responded to some of those disclosures by warning that legal action could be considered when individuals engage in activities that cause harm to customers. The company also stated that it may cooperate with law enforcement agencies when necessary.

In comments provided following the publication of the VS Code findings, Microsoft emphasized the role independent researchers play in improving product security. The company stated that it remains committed to evaluating reported issues, coordinating engineering responses, and delivering mitigations intended to protect customers.

A subsequent statement from Microsoft indicated that the issue had been mitigated within its services and that users were not required to take additional action.

The developer-focused platforms remain attractive targets because authentication tokens can provide access to source code repositories, development environments, and organizational assets. Security teams generally recommend reviewing unexpected links carefully, limiting unnecessary permissions, monitoring account activity, and using strong authentication controls to reduce the likelihood of unauthorized access.

REMUS Infostealer Reveals the Growing Sophistication of MaaS Platforms


Cybercrime is increasingly becoming an industrialized business as infostealer operations adopt the structure, speed, and feature-based development cycles of legitimate software platforms. The emergence of REMUS, as well as the development cycles associated with it, mark another shift in the industrialization of cybercrime. 


Flare researchers examined 128 underground posts published from February to May 2026, and observed the malware's rapid evolution into a full-scale malware-as-a-service ecosystem designed to facilitate operational scalability and persistent account compromise over a period of five years. This initial effort focused on harvesting saved credentials and collecting browser information, but later expanded into hijacking sessions, targeting password managers, abuse of restore tokens, and automated Telegram delivery methods, reflecting a deliberate shift toward long-term access theft rather than simple credential extraction. 

By combining REMUS's rapid updates with improved operator visibility and modular deployment capabilities, it has become apparent that REMUS will not only be used as a malware payload, but also as a commercially managed cybercrime platform aimed at supporting broader distribution, easier affiliate adoption, and increasingly resilient post-compromise operations. 

There has been an overall transformation within the underground infostealer economy that has led to operations such as REMUS maturing rapidly, where malware distribution has evolved into a highly structured commercial ecosystem, characterized by defined supply chains, subscription-based access models, dedicated log brokers and affiliate operators. 

Information theft is no longer considered as an isolated malware family but rather as the foundation for many layers of financial-motivated cybercrime. They are now playing a greater role than just stealing credentials, serving as an entry point for larger compromise operations, which include the deployment of ransomware and unauthorized access to corporate networks. 

Recent DBIR assessments indicate that credentials linked to 54 percent of ransomware victims were previously disclosed through infostealer logs, and nearly 40 percent of those datasets contained corporate email accounts, indicating that harvested session data can play a valuable role in enterprise attacks. 

A type of advanced remote access Trojan called an infostealer operates silently within infected systems, gathering cookies, authentication tokens, stored passwords, fingerprints, and other telemetry from the infected system before packaging the information into standardized "stealer logs" for exfiltration. 

In turn, these logs are sold as monetizable access assets on dark web marketplaces, cybercrime forums, and encrypted Telegram channels. Operators routinely distribute free log samples as promotional materials to attract buyers and expand their criminal subscriber base, further enhancing the commercialization of the infostealer ecosystem and its scalability. 

A detailed examination of the operator's advertisements, feature announcements, support discussions, and update logs offers an exceptional chronological perspective on REMUS' evolution from a relatively lightweight credential stealer to a continuous, operationally efficient and commercially successful MaaS platform. 

Based on the activities observed between February and May 2026, a development model that closely resembles legitimate software operations was observed, where iterative features were released, customer-oriented improvements were made, and backend management improvements were improved rapidly. 

A number of early campaigns in February created a perception of REMUS as a trustworthy, accessible, and reliable stealer that specialized in stealing browser credentials, cookie theft, extracting Discord tokens, delivering logs through Telegram, and simplifying log management. 

Throughout the promotional language, the operator emphasized a commercial mindset, including advertising "24/7 support" alongside claims that the malware was "simple enough that even a child can figure it out," as well as boasting that its callback success rate was near 90 percent through the use of dedicated intermediary infrastructure and custom encryption algorithms. After entering an aggressive expansion phase in March, the operation shifted focus from data theft toward campaign administration and operator visibility in an effort to increase efficiency. 

In addition to enhanced delivery workflows, restore-token capabilities, worker tracking, duplicate-log filtering, and expanded statistics dashboards were introduced to provide affiliates with a greater understanding of failed executions and infection performance. 

April marked another strategic transition in REMUS's evolution, this time toward authentication-based session persistence and browser-side artifact collection. These changes signaled the emergence of a managed operational ecosystem rather than merely a standalone malignant binary.

SockS5 proxy integration, antivirtualization controls, gaming-platform targeting, as well as deeper password harvesting were all added to the malware. It also included IndexedDB extractions linked to browser extensions associated with the 1Password and LastPass browser extensions, and references to Bitwarden-related collection mechanisms. 

A noticeable shift occurred towards maintaining active authenticated environments through stolen session material instead of only relying on exposed credentials. Early May showed a slowdown in the addition of entirely new features as development focused on platform stability, restoring function refinements, optimizing collection, adjusting delivery schedules, and resolving bugs. It indicated that the operator was moving from rapid capability expansion to long-term operational reliability and service maturity. 

REMUS reflected a broader shift in the priorities of the underground malware economy by clearly pivoting towards session theft and authenticated access preservation as a defining characteristic of its operation. 

Information thieves in the previous generations primarily focused on obtaining usernames and passwords for later exploitation, REMUS consistently promoted browser cookies, authentication tokens, workflows to restore sessions, and proxy-assisted continuity mechanisms as central operational features of their operations. 

There were repeated references throughout the campaign to "Restore" capabilities, multi-proxy compatibility, and token recovery workflows indicating that the malware was designed specifically to maintain active authenticated environments as opposed to simply capturing credentials on its own. As modern security controls increasingly rely on multi-factor authentication, device trust verification, behavioral analytics, and risk-based login verification, this distinction has significant operational value for threat actors.

Through the use of stolen session artifacts, rather than raw credentials alone, attackers may be able to bypass many of these layered defenses without triggering immediate authentication challenges. This objective was further reinforced by repeated targeting of Discord, Steam, Riot Games, and Telegram environments, as persistent authenticated sessions within such platforms can be used to resell accounts, conduct fraud operations, abuse social engineering, and monetize access over the long term. 

As part of its session-focused development, REMUS has demonstrated a growing interest in browser-based password management systems as well. As of April 2026, the operator has implemented collection capabilities associated with Bitwarden, 1Password, LastPass, and IndexedDB-based browser storage mechanisms commonly used to retain locally authenticated data by modern extensions and web applications. 

While the observed activity cannot independently confirm vault decryption or direct compromise of password-manager databases, it indicates that development priorities had expanded toward harvesting browser-side storage artifacts associated with password-management workflows, although there is no independent confirmation of either. 

In addition, the campaign infrastructure itself displayed a high degree of operational maturity. Throughout the deployment cycle, the operator maintained a steady cadence of versioned releases, troubleshooting refinements, feature additions, bug remediation, statistics enhancements, and backend management improvements. These practices closely resemble legitimate software maintenance practices.

Throughout the report, references to worker management, log categorization systems, infection visibility dashboards, and loader monitoring were made, implying a structured multi-role environment, where development, deployment, infrastructure management, and monetization functions were increasingly segmented. These organizational models are similar to the organizational models found in mature malware-as-a-service ecosystems today. 

REMUS illustrates how modern infostealer campaigns have evolved from opportunistic credential theft to scalable, persistent, and monetizable platforms that enable access. As a result of the rapid development cycle, a focus on authenticated session continuity, and an increasing interest in browser-based authentication ecosystems, cybercrime has experienced a broad shift, demonstrating the increasing value of stolen access in the cybercrime landscape. 

A reminder to defenders that password protection alone is not sufficient to protect against threats increasingly engineered to exploit trusted sessions, browser storage artifacts, and post-authentication workflows.

In the near future, organizations will face increased pressure to strengthen session monitoring practices, token invalidation practices, endpoint visibility, browser hardening, and anomaly-based access controls as MaaS operations continue to adopt the speed, structure, and operational discipline commonly associated with legitimate software companies. 

There is less significance to the evolution observed in REMUS with regard to any single malware capability than it has in relation to the emergence of a professional and commercialized cyber intrusion ecosystem.