A critical unauthenticated remote code execution vulnerability (CVE-2026-58138) in the Orkes Conductor workflow platform is being actively exploited in the wild. The flaw affects versions before 3.30.2 and carries a CVSS v3.1 score of 9.8 and a CVSS v4 score of 9.3. Fortinet researchers confirmed the exploitation, urging immediate patching to version 3.30.2 or later.
The ShinyHunters extortion gang breached the data leak site operated by the Clop ransomware gang on September 19, 2026. The intrusion resulted in the defacement of Clop's Tor-based leak portal and alleged exfiltration of server data and private keys used to maintain the onion service. The incident marks an escalation of tensions between two ransomware extortion groups and has disrupted Clop's leak site operations.
A proof-of-concept attack dubbed BragJack, disclosed by researcher Gal Weizman of Forever Security, exploits AI assistant integrations in Chrome, Microsoft Edge, Opera Neon, Perplexity Comet, and Claude in Chrome. Using a single malicious extension, the Prompt Forcing technique can hijack these assistants to inject instructions or exfiltrate data. The research earned over $20,000 in bounties and two CVE assignments (CVE-2026-XXXX, CVE-2026-YYYY). Vendors are likely reviewing or deploying mitigations.
A recent EY survey of senior AI executives indicates that organizations are deploying autonomous AI and machine learning systems at a pace that exceeds the development of corresponding process controls, oversight mechanisms, and governance frameworks. The findings highlight a growing risk of compliance challenges, uncontrolled AI behavior, and security vulnerabilities due to insufficient oversight.
Vectra AI announced the launch of Ascent, a new program designed to expand its partner strategy and address the growing complexity of security environments and the rise of AI-driven attacks. The program focuses on providing broader AI expertise, services, and security outcomes to partners and customers.
According to a Dark Reading report published on September 18, 2026, unknown threat actors allegedly used AI-driven agents to breach a Spanish organization and modify personal data. The exact attack vector, target entity, and full scope of data modification have not been independently confirmed. The incident underscores the emerging trend of AI agents being leveraged in cyber operations, though details remain unverified without further source confirmation.
Microsoft has fixed a known issue that caused Windows devices to display false alerts stating that Defender Antivirus was turned off after installing recent updates. The bug, which did not involve any security vulnerability or exploitation, led to confusion among users who may have taken unnecessary manual actions. Microsoft advises installing the latest updates to resolve the issue.
In July 2025, an unknown party re-registered a domain previously operated by a content delivery network that had been wound down years earlier. Thousands of websites, code repositories, and documentation pages continue to contain hard-coded references to hostnames beneath that domain. The new owner now controls the hostname, creating potential risks for content injection, tracking, malware distribution, or supply chain attacks against sites that still reference the former CDN infrastructure.
Microsoft is introducing a new capability in Microsoft Teams that will allow administrators to customize the list of file extensions blocked for security reasons. This feature enables organizations to block custom file extensions that are commonly associated with threats, aligning Teams security with company-specific requirements. No active exploitation has been reported; the update is a proactive security enhancement.
A new advisory from Dark Reading highlights that multi-factor authentication (MFA) is not sufficient to protect against OAuth consent abuse, where threat actors trick users into granting permissions to malicious applications, thereby obtaining access tokens that bypass MFA. The report emphasizes the need for comprehensive OAuth governance, including least-privilege scopes, consent monitoring, and rapid revocation procedures.
On September 11, 2026, the Cybersecurity and Infrastructure Security Agency (CISA) added CVE-2026-85706 to its Known Exploited Vulnerabilities (KEV) Catalog. The vulnerability affects GitLab Community Edition and GitLab Enterprise Edition and involves a path traversal flaw. CISA cited evidence of active exploitation in the wild. The inclusion triggers requirements under Binding Operational Directive 26-04 for Federal Civilian Executive Branch agencies to prioritize rapid remediation on publicly exposed assets.
Check Point has disclosed a critical vulnerability in its Security Management Server and Log Server products that allows unauthenticated attackers to execute arbitrary code as root over the network. The flaw, which affects the central management infrastructure for firewall policy and administrator access, has been patched via the LivePatch update channel. Check Point reports no active exploitation at this time, but the severity and unauthenticated nature demand immediate patching.
This week's threat roundup highlights attackers leveraging self-rewriting AI agents, over 800 patched vulnerabilities, insider SIM swap attacks, and 22 additional stories. The landscape remains challenging as adversaries reuse existing weaknesses and adopt new techniques.
Security researchers from Kaspersky and ESET have linked the Iran-aligned 'hacktivist' persona Handala Hack to a Telegram-based surveillance backdoor named HEAVYGRAM and a Delphi-based utility called CRUDEEXCLUDE. The malware supports remote command execution, system and network discovery, data and Telegram session file exfiltration, screenshot capture, DLL sideloading, and password theft. Active exploitation is reported via Telegram lures, with no official patch available; mitigation focuses on user awareness and endpoint monitoring.
Cybersecurity researchers have identified a new Android malware called RatHat, assessed to be operated by China-based threat actors. RatHat is distributed via targeted smishing and malvertising campaigns that lead to deceptive third-party download portals. The malware abuses Android Debug Bridge (ADB) to retain shell access on compromised devices even after the malicious app is uninstalled, and features an AI-powered system to navigate and control devices.
A financially motivated threat actor has been linked to the development and distribution of a JavaScript-based information stealer named PhantomRaven via the npm package registry. Researchers assess with high confidence that the malware was likely written using a large language model (LLM), based on verbose comments, placeholder code, and statistical token-analysis patterns. The campaign poses a supply chain risk to npm users and downstream consumers.
Researchers have identified a cluster of 13 npm packages distributing a previously undocumented JavaScript stealer codenamed WeaselBiscuit. The malware targets Chrome extension storage to harvest credentials, tokens, and other sensitive data. Analysis indicates functional overlaps with BeaverTail and another strain associated with the Democratic People's Republic of Korea's Contagious Interview campaign, though attribution remains tentative.
Microsoft has released fixes for a maximum-severity security flaw in Azure AI Foundry that could allow unauthorized privilege escalation over a network. The vulnerability, tracked as CVE-2026-85889, carries a CVSS score of 10.0. Microsoft states that no customer action is required as updates are applied automatically.
A vulnerability discovered by Air Security affects multiple AI coding agents, allowing repository owners to replace pinned plugin code with malicious versions. The flaw impacts Claude Code, Codex, and GitHub Copilot, with patches released for Anthropic and OpenAI products but not yet for GitHub Copilot.
CISA has added CVE-2025-39682, a Linux Kernel vulnerability involving an improper check for unusual or exceptional conditions, to its Known Exploited Vulnerabilities (KEV) Catalog based on evidence of active exploitation. The vulnerability poses significant risks, particularly to publicly exposed assets, and federal agencies must prioritize remediation under BOD 26-04.
A new malware campaign leverages search engine optimization to surface counterfeit GitHub repositories mimicking LastPass Authenticator and other well-known software. The repositories distribute a previously undocumented information stealer identified as Rapuncel, which targets credentials and other sensitive data from compromised systems.
Gyazo, a popular image-sharing platform, has confirmed a data breach in which attackers exploited a server vulnerability to steal 23.6 million user records. The breach was reported on September 18, 2026, and may involve personal data. The exact nature of the vulnerability and the specific data types remain unconfirmed.
WordPress has released security patches to address a suite of core vulnerabilities, including a flaw tracked as Click2Shell. The vulnerability allows a crafted web link, when opened by a logged-in administrator, to install a theme from the official WordPress.org directory without user interaction. Security researchers from pwn.ai reported the flaw, which can be chained with other vulnerabilities to achieve remote code execution. Patches have been issued; immediate updating is recommended.
Google has disclosed and patched a high-severity privilege escalation vulnerability in the Pixel Cellular Modem, tracked as CVE-2026-58704 (CVSS 8.0). The flaw, a permission bypass due to a logic error, has been exploited in the wild in limited, targeted attacks. Users are urged to apply the September 2026 security update immediately.
