Intelligence Feed

1mo ago · unit42

Unit 42 researchers identified a security issue in AWS AgentCore Harness where default configurations allow attackers to exfiltrate plaintext credentials from AgentCore Identity via prompt injection. The built-in shell tool, enabled by default and running as root, can access the memory space of the harness process (PID 1), where credentials are temporarily stored in plaintext during runtime. By injecting malicious commands through a support ticket, an attacker can execute reconnaissance and exfiltrate a JSON Web Token (JWT) used for downstream MCP server authentication. This stolen credential, belonging to the operator's service account (mcp-service), enables unauthorized access to sensitive data such as personally identifiable information (PII). AWS acknowledged the finding but classified it under customer responsibility, emphasizing proper scoping of allowedTools and egress filtering.

1mo ago · bleeping-computer

Two previously compromised GitHub Actions, actions-cool/issues-helper and actions-cool/maintain-one-comment, were re-enabled by their maintainer on September 16, 2026, without removing the malicious payload from May's Mini Shai-Hulud supply-chain attack. The actions continued to serve an obfuscated malicious payload in 'index.js', causing dependent workflows to execute malware that targets developer tokens, credentials, and CI/CD secrets. The repositories were re-disabled on September 25 after researchers at Socket raised the alarm. Developers are advised to remove or pin these actions and rotate potentially exposed secrets.

1mo ago · talos

The article is a commentary on social engineering tactics targeting cybersecurity professionals through fake consultancy offers and job scams, emphasizing the importance of trust in the industry. It does not describe a specific malware, vulnerability, or attack campaign with technical indicators. The piece also promotes a new open-source toolkit, CAIRN, and includes general security news summaries without detailed technical analysis or IoCs tied to a single threat event.

1mo ago · hacker-news

CISA has added two vulnerabilities to its Known Exploited Vulnerabilities catalog due to active in-the-wild exploitation. CVE-2026-65660 is a code injection flaw in Microsoft Office SharePoint that allows authenticated attackers to achieve remote code execution, which Microsoft initially classified as a spoofing issue. The second vulnerability, CVE-2026-67279, affects MikroTik RouterOS and enables unauthenticated attackers to open a session channel, which when combined with CVE-2026-86060 (an argument injection flaw), forms the 'MikroTrick' exploit chain allowing full administrative takeover of vulnerable routers without credentials.

1mo ago · hacker-news

A high-severity CSRF vulnerability in Elementor Website Builder WordPress plugin versions 4.3.0 and 4.3.1 allows unauthenticated attackers to take over WordPress sites by tricking an authenticated administrator into clicking a crafted link. The flaw bypasses CSRF protection for cookie-authenticated REST API requests when the string 'elementor/v1/events/' appears in the request URI, enabling actions like creating rogue administrator accounts. The vulnerability affects over 2 million sites and has been patched in version 4.3.2.

1mo ago · bleeping-computer

Kiteworks has issued a precautionary advisory urging customers to shut down their servers for a six-hour window due to credible threat intelligence from federal authorities indicating a potential imminent cyberattack. While no confirmed breach or exploitation has been identified, the company suspects possible zero-day attacks targeting its secure file-sharing systems. The warning is preventative, with all known vulnerabilities already patched in the latest version (9.5.1). The Clop extortion gang is mentioned as a potential threat actor due to its history of targeting similar platforms.

1mo ago · bleeping-computer

The Clop ransomware gang's data leak site was compromised by the ShinyHunters extortion group through an unpatched path traversal vulnerability in Grav CMS version 1.7.43. ShinyHunters exploited the flaw to upload malicious files, deface the site, and claim theft of source code, plugins, server logs, and Tor private keys, subsequently issuing a ransom demand. Grav CMS confirmed the vulnerability, tracked as CVE-2026-42608, resides in the core of Grav and was fixed in Grav 2.0 but not backported to the 1.7 branch until version 1.7.53.4 was released following disclosure.

1mo ago · bleeping-computer

The Cybersecurity and Infrastructure Security Agency (CISA) has added four vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog, warning that they are actively exploited in attacks. These include a critical authentication bypass flaw (CVE-2026-5430) in WSO2 products, an incorrect authorization vulnerability (CVE-2026-71362) in Adobe Commerce and Magento, a high-severity code injection flaw (CVE-2026-65660) in Microsoft SharePoint, and a medium-severity SSH state-machine bypass (CVE-2026-67279) in Mikrotik RouterOS. Security firm watchTowr observed exploitation attempts against WSO2 using forged JWT tokens, and Sansec reported active exploitation of the Adobe Commerce flaw in the wild. Federal agencies are required to patch these flaws by September 27–28, 2026.

1mo ago · bleeping-computer

A cross-site request forgery (CSRF) vulnerability in Elementor plugin versions 4.3.0 and 4.3.1 for WordPress allows unauthenticated attackers to create administrator accounts by tricking a logged-in administrator into opening a malicious link. The flaw exists in the Editor Events module, which bypasses WordPress REST nonce validation when the request URI contains the 'elementor/v1/events/' path, enabling attackers to append this path via query parameters to trigger unauthorized REST API actions. The vulnerability was reported by security firm Patchstack and patched in version 4.3.2, with no CVE assigned at the time of reporting.

1mo ago · hacker-news

Cryptocurrency exchange Bitget suffered a $351.6 million theft from its hot and warm wallets, attributed to suspected North Korean threat actors. The attackers compromised a critical backend system, spoofed transaction data, and triggered unauthorized fund transfers. While customer balances remain intact and cold wallets were unaffected, withdrawals have been suspended during a security review. Bitget has engaged Mandiant and SlowMist for investigation and is collaborating with blockchain foundations to freeze hacker-controlled wallet addresses. The attack pattern aligns with known North Korean hacking groups based on IP behavior and on-chain analysis.

1mo ago · hacker-news

A new variant of the PamStealer macOS malware has been identified, featuring live command-and-control (C2) payload decryption and multi-layer persistence mechanisms. The malware is distributed via a fake cryptocurrency wallet website (wavel[.]app), which delivers a malicious disk image containing a compiled AppleScript that executes a JXA dropper. The dropper initiates a key exchange with the C2 server using X25519 to decrypt the payload, preventing static analysis. The malware employs four persistence methods, including LaunchAgent, shell hooks, and Git hooks, and ultimately deploys a Swift-based stealer to harvest credentials, browser data, keychain items, and system metadata.

1mo ago · hacker-news

Two compromised GitHub Actions, 'actions-cool/issues-helper' and 'actions-cool/maintain-one-comment', were reactivated on September 16, 2026, after being previously disabled due to their involvement in the Mini Shai-Hulud supply chain attack campaign. The repositories resumed serving malicious code that had been introduced on May 18, 2026, allowing credential harvesting from CI/CD pipelines without any new attacker action. The malicious payloads were re-downloaded and executed by workflows referencing the affected version tags, highlighting the risk of mutable version tags in supply chain security. The incident is linked to the Mini Shai-Hulud activity cluster, which also targeted npm packages under the @antv ecosystem.

1mo ago · socket-dev

The compromised GitHub Actions repositories 'actions-cool/issues-helper' and 'actions-cool/maintain-one-comment' were re-enabled on September 16, 2026, while still hosting malicious code from the May 2026 Mini Shai-Hulud campaign. This reactivation allowed the malicious payload to execute in downstream workflows that referenced the actions by mutable tags, affecting an estimated 15,000+ repositories. The attack resumed without any new exploit or infrastructure, as the malicious tags were never cleaned. Workflows referencing these actions by tag instead of pinned commit SHA began executing the obfuscated payload, which installs the Bun runtime and runs a malicious script, potentially exfiltrating secrets and gaining unauthorized access.

1mo ago · hacker-news

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added two critical vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog due to evidence of active exploitation. CVE-2026-5430 is a path traversal flaw in WSO2 API Control Plane, API Manager, Traffic Manager, and Universal Gateway that enables unrestricted file upload and remote code execution. CVE-2026-71362 is an incorrect authorization vulnerability in Adobe Commerce and Magento that allows attackers to escalate privileges and access sensitive customer data without user interaction. Exploitation of both vulnerabilities has been observed in the wild, with attacks detected as early as September 10, 2026.

1mo ago · hacker-news

Cloudflare patched a vulnerability in its Containers and Sandboxes services that allowed a customer's container to read leftover disk data from previously deleted containers on the same server. The issue stemmed from thin-provisioned disks that were not properly wiped before reallocation, enabling data remnants—including SQLite databases, .env files, and browser profiles—to be recovered. The flaw was reported by researcher Oren Yomtov via Cloudflare's bug bounty program and was fixed by re-enabling block wiping and retiring all running container disks and caches. Cloudflare found no evidence of exploitation beyond authorized testing.

1mo ago · hacker-news

A pre-authentication SQL injection vulnerability in Roundcube Webmail, tracked as CVE-2026-48842, is being actively exploited in the wild. The flaw exists in the virtuser_query plugin of versions 1.6.x before 1.6.16 and 1.7.x before 1.7.1, allowing unauthenticated attackers to inject arbitrary SQL and potentially access mail account credentials and stored messages. The Canadian Centre for Cyber Security and SentinelOne have confirmed active exploitation, though specific threat actor details remain limited. Patches were released in May 2026, but over 500,000 Roundcube instances remain internet-exposed, with at least 10 identified as vulnerable as of late September 2026.

1mo ago · bleeping-computer

Cryptocurrency exchange Bitget suffered a breach in which $351.6 million was stolen from its hot and warm wallets. The attack is attributed to suspected North Korean hackers who compromised a critical backend system within Bitget's wallet infrastructure, enabling them to forge transaction data and trigger fund transfers. The breach affected multiple blockchain networks including Ethereum, XRP Ledger, Arbitrum, Avalanche, Optimism, BSC, and Base, with XRP suffering the largest single-chain loss. Bitget has suspended withdrawals pending investigation and confirmed that its cold wallets and self-custodial Bitget Wallet were unaffected. The losses will be covered by the User Protection Fund.

1mo ago · bleeping-computer

A new variant of the MacSync info-stealing malware targeting macOS systems has evolved to use public iCloud calendar events as a delivery mechanism for new payloads. Distributed via social engineering and fake applications such as a counterfeit crypto wallet called Toria, the malware retrieves commands from the description field of a public iCloud calendar, which are then executed in the macOS zsh shell. The infection chain leads to the deployment of a backdoor module written in Objective-C that masquerades as Finder, establishes persistence via LaunchAgent, .zshrc modifications, and Git hooks, and can execute remote AppleScripts, deploy malicious browser extensions, and exfiltrate system data.

1mo ago · datadog-security-labs

A remote code execution (RCE) vulnerability, tracked as GHSA-632h-h47v-g4x4, was discovered in OpenCode versions 1.14.30 through 1.18.21. The flaw exists in the /global/upgrade API endpoint, where a content-type confusion allows attackers to exploit a code injection vulnerability by submitting a malicious npm package via a cross-origin POST request. The attack can be triggered when a user visits a malicious webpage while running an unsecured OpenCode instance, leading to execution of preinstall scripts from the attacker-controlled package. The vulnerability was patched in OpenCode 1.18.22 by restricting the upgrade target to valid semantic versions and enforcing proper content-type validation.

1mo ago · hacker-news

A malicious Android spyware dubbed Corp MDM is targeting logistics firms by distributing fake Google Play pages impersonating CEVA and TKW Logistics. The malware, delivered as a trojanized APK with package name 'com.corp.mdm', steals newly received SMS messages, redirects calls, and maintains a hidden foreground service. It communicates with a command-and-control server at 69.55.61.82 via HTTP, exfiltrating SMS content and device identifiers while supporting remote commands such as call forwarding and self-destruction. The campaign is suspected to be financially motivated, with links to a Russian-Armenian threat actor operating a phishing-as-a-service platform called Global Profit.

1mo ago · hacker-news

A threat campaign dubbed ClickFix is compromising legitimate Ukrainian business websites to serve fake Cloudflare verification pages that trick users into executing a malicious command. The command downloads an MSI installer delivering Psychedelic Stealer, a previously undocumented information stealer that exfiltrates browser credentials, cryptocurrency wallets, and account tokens. The attackers also use a lure management panel hosted on uasputnik.com, and the malware establishes persistence via scheduled tasks and communicates with C2 servers. A second malware chain delivers RemotePanel, a remote access tool, and BoundSiphon, a .NET stealer, using similar social engineering lures.

1mo ago · hacker-news

The domain third-party[.]com, historically used as a documentation placeholder in code and technical writing, has been registered by an attacker and is now serving a ClickFix social engineering lure targeting Windows users. When accessed from Windows, the site poisons the clipboard with a malicious PowerShell command intended for execution via the Run dialog, while showing a benign or unsupported message to other platforms like macOS. The domain is referenced in over 1,700 public GitHub repositories, including AI agent skills and API documentation, amplifying exposure. Additionally, two other placeholder domains—yoursite[.]com and your-domain[.]com—are actively serving scams and scareware, particularly targeting macOS users with fake security alerts and fraudulent investment offers.

1mo ago · hacker-news

The article details multiple active cyber threats, including the RemControl Android banking trojan targeting users in Western Europe, the Middle East, and Canada via fake Google Play Store pages distributed through Meta ads. The malware abuses Android Accessibility Services to perform screen streaming, keystroke logging, and remote control, with command-and-control infrastructure dynamically resolved through encrypted Telegram dead-drops. Phishing overlays and operator panels show signs of AI-assisted development, and Russian-language code comments suggest Russian-speaking involvement. The campaign shares infrastructure and tactics with the Medusa UNKN affiliate botnet, indicating a coordinated threat operation.

1mo ago · hacker-news

A security researcher, Rasmus Moorats, identified two unpatched vulnerabilities in OnePlus devices running OxygenOS that can be chained to allow a locally installed Android app with no permissions to gain full root access. The first flaw is in the AtlasService, which runs as root and accepts unvalidated input from any app, enabling command injection into a restricted root environment. The second flaw leverages the olc2 hardware helper service, which executes arbitrary shell commands if the caller has root—achieved via the first flaw—granting full system-level control. As of September 24, 2026, no fix or CVE had been issued by OnePlus, and the company warned the researcher against public disclosure, citing legal consequences.

1mo ago · bleeping-computer

A critical pre-authenticated SQL injection vulnerability in Roundcube Webmail, tracked as CVE-2026-48842, is now being actively exploited in the wild. The flaw exists in the virtuser_query plugin and allows unauthenticated attackers to bypass authentication, execute malicious database commands, and steal data without user interaction. The Canadian Centre for Cyber Security has issued an updated advisory warning of ongoing exploitation, urging administrators to update to patched versions 1.6.16 or 1.7.1, or disable the vulnerable plugin if immediate patching is not possible. Roundcube instances have been frequent targets in the past, including by state-backed groups such as APT28 and TA473.

1mo ago · hacker-news

ClickFix is a social engineering-based initial access technique that manipulates users into pasting malicious commands into trusted system interfaces, bypassing traditional security controls. The attack leverages compromised websites, often WordPress-based, to serve fake verification pages that trigger clipboard manipulation and user-driven command execution. Infrastructure is resilient, using blockchain (Polygon) and Telegram/Steam for dynamic domain resolution, enabling rapid rotation and evasion of blocklists. Payloads are fingerprint-gated, delivering malware like Vidar Stealer only to specific victims based on hardware and account identifiers, making sandbox analysis unreliable.

1mo ago · bleeping-computer

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has warned that ransomware gangs are actively exploiting a critical authentication bypass vulnerability in JetBrains TeamCity, tracked as CVE-2026-63077. This flaw allows unauthenticated attackers to execute arbitrary operating system commands via the TeamCity agent polling protocol, potentially compromising CI/CD pipelines, stealing credentials, and modifying server state. CISA added the vulnerability to its Known Exploited Vulnerabilities catalog and mandated federal agencies to remediate within three days. Although patching is available, hundreds of internet-exposed TeamCity servers remain unpatched, making them attractive targets for ransomware and state-sponsored actors.

1mo ago · hacker-news

Threat actors are actively exploiting CVE-2026-87902, a critical vulnerability in WordPress that allows unauthenticated remote code execution (RCE) under specific conditions. Exploitation requires the active theme to have a directory starting with 'page-' and a readable local PHP file such as 'pearcmd.php' on the server. Attackers are leveraging the vulnerability to write PHP web shells to temporary directories and are retrieving payloads from a GitHub repository. Multiple IP addresses have been observed conducting exploitation attempts, with the first recorded just hours after the patch was released.

1mo ago · hacker-news

The UNK_CondorFiltration campaign, leveraging the TeamFiltration framework, conducted brute-force attacks against Microsoft 365 accounts across 28 tenants, primarily targeting Chilean financial and retail organizations. The attackers exploited unmanaged service accounts with default or unrotated passwords and no multi-factor authentication, compromising seven accounts. Six of the seven breaches occurred within seven minutes of each other, suggesting shared credentials. Post-compromise, the threat actor used German VPN nodes to access corporate resources including Azure Portal, SharePoint Online, and Microsoft Graph API.

1mo ago · socket-dev

A malicious Firefox extension named 'pdf-para-texto@extensao.local' poses as a PDF identity verifier to target Portuguese- and Spanish-speaking users, enabling automated Google account takeover. The extension avoids static detection by shipping without hardcoded malicious code and instead fetches its payload post-installation from attacker-controlled infrastructure. It steals Google session cookies (oauth_token) and can silently reset passwords during Google's sign-in flow, providing attackers with both immediate and persistent access. The malicious behavior is triggered after installation via a lookalike domain (pdf[.]gusercontent[.]com), which delivers configuration data that arms the extension with credential-theft and automation capabilities.