RovoBlast: Critical Atlassian Rovo AI Vulnerability Exposes Enterprise Data via Malicious Links Security researchers at Varonis Threat Labs have uncovered RovoBlast, a severe vulnerability in Atlassian’s Rovo AI assistant that enables attackers to extract sensitive enterprise data through a single malicious link. The flaw exploits Rovo’s handling of URL-supplied prompts, allowing threat actors to inject malicious instructions into an authenticated user’s AI session without requiring account compromise or permission bypasses. The attack leverages a parameter-to-prompt (P2P) weakness, where Rovo treats text embedded in a URL (via the `rovoChatPrompt` parameter) as a pre-filled chat instruction within a trusted session. A crafted link such as `https://home.atlassian.com/chat?rovoChatPathway=chat&rovoChatPrompt=<malicious_prompt>` can trigger the assistant to execute attacker-controlled commands when clicked by a logged-in user. Rovo, designed as an AI layer across Atlassian products (Jira, Confluence, Bitbucket) and external platforms (Slack, Microsoft 365, Google Workspace, databases, and web resources), operates under the permissions of the authenticated user. This means malicious prompts can search, summarize, or exfiltrate data while appearing as legitimate AI-assisted workflows. The ResearchAgent feature is particularly concerning, as it can perform multi-step research, navigate websites, and potentially chain actions to retrieve and transmit sensitive information all from a single prompt. Unlike traditional prompt-injection attacks, RovoBlast does not require elaborate jailbreaks or repeated inputs, making it harder for defenders to detect. The vulnerability underscores the risks of AI assistants processing untrusted content while holding broad data access. Organizations are advised to restrict Rovo’s integrations, disable unnecessary automation features, and monitor AI logs for unusual behavior to mitigate exposure. The discovery highlights a growing security challenge: AI assistants with expansive permissions and external connectivity can become low-friction vectors for data exfiltration.

Feral Wolf Ransomware Campaign Exploits Exposed Business Software in Russian Organizations A recent ransomware campaign by the threat group Feral Wolf has targeted Russian organizations across retail, construction, manufacturing, and IT sectors from May to August 2026, leveraging exposed business software and weak server configurations to infiltrate corporate networks. The attackers deployed GenieLocker ransomware after gaining access through vulnerable systems, demonstrating how a single overlooked internet-facing service can lead to a full-scale breach. Security firm BI.ZONE uncovered the intrusions, tracing attack paths through unpatched Atlassian Confluence installations, poorly secured 1C:Enterprise clusters, and contractor environments. In one case, Feral Wolf exploited CVE-2023-22515 on a publicly accessible Confluence server running in a Docker container, creating an admin account to establish a foothold. From there, they moved laterally by compromising a PostgreSQL service with weak credentials, escalating access to the underlying host. The group also abused misconfigured 1C:Enterprise server clusters, exploiting unprotected management services and debug modes to execute arbitrary commands. Their tactics included custom backdoors (MQTTDoor and MatrixDoor), which communicated via MQTT and Matrix protocols, blending malicious traffic with legitimate network activity. Additionally, they used proxy tools to tunnel traffic through Remote Desktop Protocol (RDP) sessions, further evading detection. Post-compromise, Feral Wolf employed credential theft techniques, extracting memory from Windows systems using legitimate utilities, and attempted to erase forensic evidence with PowerShell scripts. The campaign highlights the risks of unpatched collaboration platforms, exposed database services, and weak authentication controls, emphasizing the need for network segmentation, log monitoring, and rapid patching to disrupt multi-stage ransomware attacks. Indicators of Compromise (IoCs) associated with the campaign include: - IPs: `45.151.45[.]31` (Confluence intrusion), `46.166.79[.]31` (PostgreSQL server) - Domains: `broker.hivemq[.]com` (MQTT broker), `meet.element[.]tw` (Matrix homeserver) - Tools: `PrintSpoofer64.exe`, `nyx.ps1`, `fscan`, and GenieLocker ransomware samples - Hashes: Multiple SHA-256 values for backdoors and malware (e.g., `MQTTDoor`, `MatrixDoor`) The investigation underscores that ransomware incidents often stem from chained vulnerabilities, combining known exploits, credential abuse, and covert communication channels to evade detection before encryption.

Critical OS Command Injection Flaw in Atlassian Bamboo Puts CI/CD Pipelines at Risk Atlassian disclosed a critical security vulnerability (CVE-2026-21571) in Bamboo Data Center and Server, allowing remote attackers to execute arbitrary operating system commands. The flaw, assigned a CVSS score of 9.4, was published on April 21, 2026, as part of Atlassian’s monthly Security Bulletin. The vulnerability affects multiple versions of Bamboo, including 9.6.0, 10.0.0, 10.1.0, 10.2.0, 11.0.0, 11.1.0, 12.0.0, and 12.1.0. It stems from a third-party dependency but remains classified as critical due to its potential impact. Exploitation requires low-level authentication and no user interaction, making it a high-risk threat for enterprise environments. Successful attacks could enable threat actors to inject malicious code into CI/CD pipelines, compromising software supply chains, accessing sensitive data, or disrupting system operations. Given Bamboo’s role in automating build and deployment workflows, unpatched systems pose a significant risk to development environments. Atlassian has released patched versions (12.1.6 (LTS), 10.2.18 (LTS), and 9.6.25) to mitigate the flaw. Organizations unable to upgrade immediately are advised to review Atlassian’s Vulnerability Disclosure Portal for mitigation steps, including monitoring authentication logs and auditing CI/CD pipelines for unauthorized changes. The April 2026 Security Bulletin also addressed 37 additional vulnerabilities, including a CVSS 10.0 cross-site scripting flaw and a remote code execution issue in other Atlassian products like Jira, Confluence, and Bitbucket.

Critical Stored XSS Vulnerability in Atlassian Jira Enables Full Organization Takeover Security researchers at SnapSec recently disclosed a severe stored Cross-Site Scripting (XSS) vulnerability in Atlassian’s Jira Work Management, a widely used platform for project tracking and task management. The flaw, stemming from inadequate input validation in a low-risk settings menu, allows attackers with limited administrative permissions to execute a full organization takeover. ### Vulnerability Details The issue resides in Jira’s custom priority settings, where administrators can define task importance levels (e.g., high, medium, low). While editing these priorities, users can specify an Icon URL a field that, if manipulated, could inject malicious JavaScript. Researchers demonstrated that a Product Admin a role with restricted but sufficient permissions could embed a payload in the URL (e.g., `https://google.com?name=</script><script>alert(0)</script>`). Due to missing backend validation and output encoding, the script was stored in the database and executed when a Super Admin accessed the priorities configuration page. ### Exploitation & Impact The attack leverages stored XSS, meaning no victim interaction (e.g., clicking a link) is required. Once a Super Admin loads the compromised page, the malicious script executes in their browser, operating within a highly privileged administrative context. In SnapSec’s proof-of-concept, the payload silently sent a system invitation to an attacker-controlled account, granting them full access to Jira, Confluence, and other Atlassian products. This enabled unauthorized project creation, modification, and deletion effectively seizing control of the entire organization. ### Key Takeaways - The vulnerability exposes a critical gap in input validation, even in mature enterprise platforms. - Partially privileged roles (e.g., Product Admins) can escalate to full administrative control if access controls are not rigorously audited. - The incident underscores the need for strict backend validation and output encoding across all configuration panels, regardless of perceived risk. Atlassian has since addressed the flaw, but the discovery serves as a reminder that overlooked administrative features can become high-impact attack vectors.

Atlassian Patches High-Severity RCE Vulnerability in Bamboo Data Center Atlassian has addressed a high-severity remote code execution (RCE) vulnerability, CVE-2026-21570, affecting its Bamboo Data Center application. The flaw, discovered internally through Atlassian’s security auditing program, poses significant risks to enterprise CI/CD environments, where Bamboo serves as a critical hub for automated builds, testing, and deployment. With a CVSS 4.0 score of 8.6, the vulnerability allows authenticated attackers with elevated privileges to execute arbitrary code remotely on affected servers. Exploitation could lead to full system compromise, enabling threat actors to manipulate source code, exfiltrate sensitive build secrets, or disrupt software development operations potentially facilitating devastating supply chain attacks. The flaw impacts multiple Bamboo Data Center versions, including: - 9.6.x (9.6.0–9.6.23) - 10.0.0, 10.1.0, 10.2.0 - 11.0.0, 11.1.0 - 12.x (12.0.0–12.1.2) Atlassian has released patches to mitigate the issue, urging administrators to upgrade immediately: - 9.6.x → 9.6.24 or later - 10.2.x → 10.2.16 - 12.1.x → 12.1.3 or later Patched versions are available via the Atlassian download center. Organizations running affected deployments are advised to apply updates to secure their build infrastructure.

ShinyHunters-Linked Cybercrime Campaign Targets Over 100 Major Organizations A recent cybercrime campaign attributed to the ShinyHunters group has targeted at least 100 organizations across multiple sectors, including software, finance, healthcare, and energy, according to cybersecurity firm Silent Push. Over the past 30 days, threat actors registered fake domains impersonating high-profile companies such as Atlassian, Adyen, Canva, Epic Games, HubSpot, Moderna, ZoomInfo, GameStop, WeWork, Halliburton, Sonos, and Telstra. The attackers employed voice phishing (vishing) tactics to compromise single sign-on (SSO) accounts, particularly those using Okta and other identity platforms. Using specialized phishing kits, they intercepted credentials and manipulated victims into bypassing multi-factor authentication (MFA) by convincing them to approve push notifications or submit one-time passcodes (OTPs). Okta described the attacks as involving real-time session orchestration, where threat actors guided victims through the authentication process via verbal instructions. While Silent Push identified the infrastructure used in the campaign, it remains unclear whether the attacks successfully breached any systems. However, ShinyHunters has claimed responsibility for data breaches at companies like Betterment, Crunchbase, and SoundCloud, all of which confirmed incidents. The group allegedly stole millions of records from these organizations as part of the Okta SSO vishing campaign. Silent Push attributes the campaign to Scattered LAPSUS$ Hunters, a collective formed last year by members of Lapsus$, Scattered Spider, and ShinyHunters, based on observed tactics, techniques, and procedures (TTPs). The incident follows recent warnings from Google and others about rising vishing and phishing attacks targeting identity platforms.

Feral Wolf Ransomware Group Exploits Confluence Flaws to Target Russian Organizations The Feral Wolf ransomware group has launched a campaign against Russian organizations, exploiting exposed Atlassian Confluence servers, misconfigured 1C:Enterprise systems, and compromised contractor credentials to deploy GenieLocker ransomware. The attacks leverage CVE-2023-22515, a critical Confluence vulnerability, as an initial entry point, followed by privilege escalation using CVE-2021-4034 and CVE-2026-31431. Feral Wolf, a financially motivated threat actor, conducts extensive reconnaissance before establishing persistent access. The group employs anti-forensic tools to erase logs and uses MQTT, Matrix, and RDP-based tunnels to obscure command-and-control traffic. Weakly secured 1C configurations and stolen credentials further enable lateral movement, leading to data encryption, operational disruption, and prolonged recovery efforts. The campaign underscores the risks of unpatched software and exposed administrative interfaces, particularly in enterprise environments. Organizations are advised to prioritize patching, enforce strong authentication, and restrict direct internet access to critical systems. The findings were reported by Smarter MSP.

Atlassian Confluence Data Center and Server versions were affected by a critical vulnerability identified as CVE-2023-22527, enabling threat actors to exploit the flaw for cryptomining campaigns. Due to the template injection vulnerability, remote attackers could execute arbitrary code, leading to unauthorized cryptocurrency mining using the organization's resources. This activity not only utilized the compromised infrastructure for mining but also had the potential to disrupt operations and financials through resource exhaustion and increase in operational costs. Atlassian released patches to address the issue, however, systems not updated remained at risk.

Atlassian reveals a data leak that was brought on by the theft of employee login information that was then utilized to obtain data from a third-party vendor. More than 13,200 entries make up the employee file that was uploaded online, and a brief inspection of the file suggests that it contains data on many current employees, including names, email addresses, work departments, and other details. The threat actors obtained information from a third-party vendor using the employee login credentials they had stolen. The business emphasized that the event had no impact on consumer or network data. The business acknowledged the data breach and disclosed that Envoy, a startup that offers workplace management services to the Australian software giant, was the source of the leaked data.

Atlassian warned its customers that multiple threat groups are exploiting a Confluence Server zero-day vulnerability in its servers. Any unauthenticated attackers can target its Confluence Server and Data Center by a critical vulnerability that can be exploited for remote code execution. The company advised its users have been advised to prevent access to their Confluence servers from the internet, or simply disable these instances, as all supported versions of Confluence Server and Data Center are affected. However, Atlassian expects fixes to become available soon.

Atlassian discovered a vulnerability in its Confluence Server which they need to patch to remedy a Critical-rated flaw. Confluence Server Webwork OGNL injection vulnerability could allow an authenticated user, or unauthenticated user, to execute arbitrary code on a Confluence Server or Data Center instance. However, Atlassian's own Confluence Cloud was patched but other hosted Confluence offerings might be vulnerable.

Atlassian revealed that unidentified hackers gained access to a vast quantity of data from its group chat service HipChat by breaking into a cloud server owned by the business. Although Atlassian did not disclose the identity of the prominent third-party software library that was utilised by its HipChat.com service, the business claims that attackers took advantage of a weakness in the library. The business issued instructions on how to reset passwords to all users whose accounts were connected to HipChat and, as a precaution, invalidated the passwords on those accounts. The organisation claims that although hashed passwords, email addresses, and names were accessible to hackers, no financial information was revealed.