Intelligence Feed

3d ago · datadog-security-labs

CVE-2026-111497 is a critical out-of-bounds write vulnerability in the Google Brotli C library, affecting multiple runtimes including Node.js, Python, Bun, .NET, Java brotli4j, Envoy, and Istio. The flaw allows memory corruption during Brotli decompression, enabling attackers to achieve denial of service, heap memory reads, and potentially remote code execution. Exploitation has been confirmed in the wild, with CISA listing it in its known exploited vulnerabilities catalog. The vulnerability stems from a boolean return value ambiguity in DecodeBlockTypeAndLength(), leading to unbounded memory writes and exploitable state corruption.

1h ago · socket-dev

The article discusses two major security issues: the compromise of eMudhra intermediate CA private keys due to a flawed key-generation process, leading to weak 8-bit keys being used instead of 384-bit keys, and a memory corruption vulnerability (CVE-2026-111497) in the Brotli decompression library. The eMudhra incident allowed potential certificate forgery, with trust chains extending through roots like Sectigo’s AAA Certificate Services, while the Brotli flaw affected multiple runtimes and applications including Node.js, Bun, Python (via requests), and .NET. Despite patches being released, the widespread bundling of vulnerable libraries and certificate stores in applications means updates must be manually propagated, leaving many systems exposed.

1h ago · socket-dev

A malicious version of the Claude Code extension, 2.1.313, was distributed through Cursor's marketplace, potentially affecting up to 5,000 users. The compromised extension included code that added an SSH authorized key, attempted to establish a reverse SSH tunnel, and installed persistence mechanisms. While initial reports suggested distribution via Open VSX or npm, investigation found no evidence of the malicious version being published to those registries. Cursor confirmed the compromise, removed the extension, rebuilt its marketplace with rotated keys, and worked to take down the attacker's infrastructure.

1h ago · wiz

A critical out-of-bounds write vulnerability, CVE-2026-111497 (Broccoli2Shell), has been identified in the Brotli decompression library, affecting multiple widely used technologies including Node.js, Bun, Envoy Proxy, and Python packages. The flaw allows unauthenticated remote attackers to trigger memory corruption, disclose sensitive information, crash processes, or potentially achieve remote code execution by supplying maliciously crafted Brotli-compressed data. Cloud environments are especially at risk, with Wiz reporting that 91% of cloud environments have vulnerable resources and 82% have publicly exposed Node.js instances, highlighting widespread exposure across containerized applications, Kubernetes clusters, and serverless workloads.

1h ago · wiz

The article details ongoing exploitation attempts against the recently disclosed PHP vulnerabilities CVE-2026-109618 and CVE-2026-109619, collectively referred to as PHP2Shell. Wiz Research observed multiple attack variants using different exploit kits, primarily targeting memory exhaustion and session poisoning, though no complete end-to-end exploitation was confirmed in their telemetry. Attackers used techniques such as sending oversized multipart requests to trigger memory exhaustion and attempted to leak process memory, but these stages failed due to differences in PHP configurations and deployment environments. The report also notes that some IP addresses successfully exploited honeypots, contributing to CISA's inclusion of PHP2Shell in its KEV catalog, while other tools labeled as PHP2Shell exploits actually relied on unrelated session poisoning and LFI techniques.

1h ago · wiz

A coordinated campaign dubbed 'Black October' targeted multiple AI developer platforms in October, resulting in at least nine disclosed security incidents. The attacks exploited credentials through various initial access vectors, including supply chain compromises, privilege escalation, and third-party dependencies. Threat actors focused on stealing AI model access keys to enable LLMJacking—monetizing stolen tokens via resale platforms. Wiz observed downstream abuse of compromised enterprise credentials, with attackers validating keys, probing model availability, and launching massive inference floods across services like AWS Bedrock, generating over $100k in estimated usage. Two resale hubs, linked to IPs 23.106.46.239 and 38.246.114.118, were identified as key players in the illicit LLM token economy.

1w ago · step-security

On October 19, 2026, an attacker compromised the GitHub account HassenBF, a legitimate OVHcloud contributor, and replaced binaries in five OVHcloud CLI releases (v0.11.0–v0.15.0) with a malicious version containing the GlassWorm credential stealer. The malicious binaries were live for 94 minutes, downloaded approximately 250 times, and exfiltrated sensitive files including cloud credentials, SSH keys, and AI agent tokens to an endpoint at 45.32.148.174. The attack leveraged a previously compromised developer machine via a malicious VS Code extension, and the payload was delivered through a trojanized release binary that bypassed checksum verification due to swapped hashes. The attacker used a local Go build without proper version flags, leaving a detectable fingerprint via the 'ovhcloud version' command. The malicious activity was detected by a customer using pinned SHA-256 hashes in CI, leading to rapid remediation and release of a clean, immutable, attested v0.16.0 version.

6h ago · step-security

On October 29, 2026, malicious actors published trojanized versions (2.1.312 and 2.1.313) of the official Claude Code VS Code extension to Cursor's marketplace, delivering a persistent SSH backdoor to developer machines. The compromised extension.js file contained a self-contained loader that established reverse SSH tunnels and HTTP beacons to attacker-controlled infrastructure. The attack enabled unauthorized access via a dropped SSH key labeled 'svc-telemetry' and established persistence through scheduled tasks and LaunchAgents. The malicious builds were distributed exclusively through Cursor's marketplace and did not affect official Anthropic distribution channels.

5h ago · step-security

The FBI has attributed a $6.3 billion cryptocurrency theft from Binance to North Korea's TraderTraitor threat actor group. The initial access point is suspected to be a compromised DevOps laptop, which allowed attackers to reach a service container and submit unauthorized withdrawal requests. While the exact infection method remains under investigation, the attackers exploited legitimate signing infrastructure without directly stealing private keys, using Ethereum's EIP-7702 delegation feature to authorize multiple fund sweeps. This incident highlights the risk posed by compromised developer machines with access to critical systems, especially given prior TraderTraitor activity involving malicious Python scripts sent via fake recruitment tests.

1w ago · unit42

Threat actors are actively exploiting CVE-2026-0257, a PAN-OS authentication bypass vulnerability in GlobalProtect portal and gateway components, to gain unauthorized VPN access. Exploitation has been linked to ransomware operators including Spikey Scorpius (Qilin), Tarnished Scorpius (Inc), and Settra ransomware-affiliated actors. Attackers bypass MFA, establish initial footholds, and conduct reconnaissance, lateral movement via RDP/SMB, and credential harvesting. In advanced cases, they pivot to virtualization layers, targeting vCenter and ESXi to dump domain controller credentials offline. The vulnerability was added to CISA's KEV catalog on May 13, 2026, and fixes have been available since May 2026, but unpatched systems remain at risk.

3d ago · talos

Cisco Talos identified UAT-11473, a sophisticated APT group, exploiting CVE-2026-20127, an authentication bypass vulnerability in Cisco SD-WAN Catalyst Controller, to gain unauthorized access. The actor deployed a custom malware framework named 'KeyDrift', which includes components like PolyFlame and LittleFilch, designed to intercept decrypted control traffic, extract data-plane keying material, and establish backdoors. Targeted sectors include space and satellite services, defense, aerospace, and policy research. Talos assesses with high confidence that UAT-11473's infrastructure overlaps with Volt Typhoon’s 'Silent Train' network.

2d ago · talos

Cisco Talos identified a new AI-integrated command-and-control (C2) framework named NIK C2, which enables threat actors to conduct credential theft, keystroke logging, ransomware deployment, and AI-driven data analysis directly from compromised hosts. The malware, written in Go, uses a dual-channel C2 architecture over TLS-secured WebSockets and HTTP polling, with traffic encrypted via AES-256-GCM. A key feature is its integration with OpenAI's API, allowing attackers to route prompts through victim machines using either their own or the victim's API key, thereby obscuring attribution and shifting costs to the victim. The framework supports over 52 commands, including screen-locking ransomware, lateral movement, and bootkit preparation, and has been observed in active infrastructure, confirming it as a functional threat.

2d ago · talos

Cisco Talos discovered a new Go-based Windows implant called NIK C2, which functions as a command-and-control framework integrating AI directly into the attack chain. This malware enables adversaries to dump credentials, stream keystrokes, and deploy screen-locking ransomware, while using a victim's own OpenAI API keys to fund and conceal malicious activity. The integration of AI allows real-time, context-aware payload generation on the compromised host, eliminating traditional detection windows and obscuring attacker infrastructure by leveraging legitimate API usage.

16h ago · talos

VOLTA-AUTODOXER is a sophisticated Windows infostealer that steals credentials, session tokens, cryptocurrency wallets, and other sensitive data, then enriches the stolen data using live API calls to services like Gmail, Discord, and Roblox. It uniquely sends a structured evidence bundle to third-party LLMs via OpenRouter to generate a prioritized, AI-analyzed dossier of the victim's monetizable assets and security posture. The malware uses a fallback chain of LLMs (Claude, Kimi, Gemini) to bypass model refusals and operates without attacker-owned infrastructure, exfiltrating data in encrypted ZIP files to Telegram. It is distributed via a builder model, indicating commoditization and broad accessibility to cybercriminals.

1d ago · hacker-news

Trezor has issued a warning about a suspected hardware implant in one of its Safe 5 hardware wallets purchased from the reseller CryptoBilis. A customer in Indonesia reported an additional circuit board between the main board and screen, containing a cellular module and antenna, capable of reading the recovery phrase from the screen and transmitting it over mobile networks. Trezor considers the report credible enough to advise all CryptoBilis buyers since March 2026 to move funds to a new wallet with a new recovery phrase, as physical inspection is unreliable and tampered devices may already have exfiltrated sensitive data. This incident follows Ledger's similar findings involving CryptoBilis, and investigations are ongoing without attribution to a specific threat actor.

1d ago · hacker-news

A high-severity remote code execution vulnerability in WordPress Core versions 4.7 through 7.1.3 has been actively exploited in the wild, with 62,969 attack attempts observed within 24 hours of the patch release. The flaw stems from predictable, web-accessible staging paths used during full-package core upgrades, allowing unauthenticated attackers to execute PHP code by exploiting a race condition when the WordPress installer is temporarily exposed. Attackers are polling for vulnerable paths like 'setup-config.php' to exploit the brief window during which the staging directory is accessible, prompting urgent advice for site owners to update to version 7.1.4 and monitor for suspicious activity.

1d ago · hacker-news

A critical vulnerability in TeamViewer, tracked as CVE-2026-111585, allows a malicious or compromised host to attack the viewer's machine during a remote session. The flaw stems from an integer overflow when processing screen size data, leading to insufficient memory allocation and subsequent out-of-bounds memory writes. This can result in crashes or, on certain platforms like Windows, remote code execution. The vulnerability affects TeamViewer versions prior to 15.83 across Windows, Linux, macOS, and Android, with no known in-the-wild exploitation reported at the time of disclosure.

1d ago · hacker-news

Chinese state-sponsored threat actor Vapor Panda, also tracked as Earth Estries, has been using an AI-powered penetration testing framework called ARTEX to conduct large-scale intrusions against over 200 organizations globally since October 17, 2026. The group leverages AI agents powered by ARTEX in combination with commercial LLMs to automate reconnaissance, exploitation, and lateral movement across cloud, container, Active Directory, and VPN environments. Post-compromise, the attackers deploy BLOODALCHEMY, a ShadowPad derivative, and perform credential dumping, password spraying, and data exfiltration, with confirmed breaches at nine organizations and attempted intrusions at 12 others.

23h ago · hacker-news

Researchers at Parameter discovered a critical vulnerability in ChatGPT that allowed a maliciously crafted public share link to execute attacker-controlled code within a victim's authenticated session. By exploiting a combination of features—Preview rendering and an insecure message listener from the MathLive library—attackers could bypass sandboxing and inject HTML into the trusted chatgpt.com context, enabling full account takeover. This could allow the reading of private conversations and performing any action on behalf of the victim. OpenAI patched the issue within eight days of disclosure, classifying it as P1, their highest severity level.

20h ago · hacker-news

A previously undocumented threat actor, dubbed UTA0576, has exploited a chain of three vulnerabilities in Ivanti Connect Secure (ICS) to achieve unauthenticated root-level code execution. The exploit chain combines an authentication bypass (CVE-2026-104935), a remote code execution flaw, and a local privilege escalation vulnerability (CVE-2026-111357) to gain full system access without credentials. Rather than deploying malware or web shells, the actor focuses on credential theft, Active Directory reconnaissance, and harvesting sensitive data such as administrator passwords and BitLocker recovery keys. Patches are now available for all affected versions, and federal agencies are required to remediate by specific deadlines.

1d ago · bleeping-computer

A critical vulnerability in GitLab, tracked as CVE-2026-97229, is being actively exploited in the wild. The flaw exists in the CI/CD Regular Expression Compiler and stems from an integer overflow weakness, allowing unauthenticated attackers to execute arbitrary code on vulnerable GitLab servers via a specially crafted regular expression in a CI/CD configuration. The US Cybersecurity and Infrastructure Security Agency (CISA) has added the vulnerability to its Known Exploited Vulnerabilities catalog and mandated federal agencies to patch by a specific deadline, while also urging private sector organizations to prioritize remediation. GitLab has released patches in versions 19.4.2, 19.3.5, and 19.2.8 for both Community and Enterprise Editions.

1d ago · bleeping-computer

A malicious version of the Anthropic Claude Code VS Code extension, version 2.1.313, was distributed through Cursor's extension marketplace for approximately 11 hours on October 29, 2026, potentially affecting up to 5,000 developers. The compromised extension executed malicious JavaScript code upon loading, which collected system and user information, installed an SSH backdoor, and attempted to establish a reverse tunnel to an attacker-controlled server. While coding errors in the malware prevented successful SSH backdoor activation on many systems, reconnaissance data was still exfiltrated to the C2 server at 134.209.82[.]225. Cursor removed the extension, took down its marketplace temporarily, and coordinated takedown of the attacker's infrastructure.

1d ago · bleeping-computer

Trezor has issued a warning about a suspected hardware implant in a Trezor Safe 5 device sold through its reseller CryptoBilis in Southeast Asia. An Indonesian customer reported discovering an unauthorized circuit board with a cellular module and antenna inside the device, which could potentially exfiltrate sensitive data such as recovery phrases over mobile networks. The same reseller was previously implicated in a similar Ledger hardware wallet compromise, suggesting a broader supply-chain attack. Trezor has not yet physically verified the device but advises users who purchased through CryptoBilis since March 2026 to transfer their assets to new wallets and avoid using potentially compromised devices.

1d ago · bleeping-computer

A critical use-after-free vulnerability in the Linux kernel, tracked as CVE-2026-98432, affects TCP connection handling and allows unauthenticated remote attackers to potentially execute code as root on systems with Multipath TCP (MPTCP) enabled. The flaw impacts Linux kernels from version 4.4 onward, including major distributions such as Ubuntu, RHEL, and Rocky Linux. While exploitation is complex, researchers at Anthropic demonstrated remote root code execution under specific conditions using an unreleased AI model. The vulnerability stems from a race condition in tcp_v4_syn_recv_sock() and can be triggered even without MPTCP in some cases, though exploitation has not been publicly demonstrated on standard TCP connections.

13h ago · bleeping-computer

A zero-click denial-of-service vulnerability in Android, tracked as CVE-2026-58882 and dubbed 'SMS of Death', allows attackers to remotely crash unpatched phones or disrupt cellular service by sending a specially crafted SMS message. The flaw exists in the WapPushOverSms.decodeWapPdu function, which fails to validate a header-length value, leading to a negative array size allocation that crashes the com.android.phone process. Google patched the issue in October 2026 for Android 14 through 17, but older Android 12 and 13 devices remain vulnerable. A proof-of-concept exploit has been released, enabling malicious actors to send the crafted message to port 2948, potentially causing reboot loops and blocking emergency calls.

10h ago · bleeping-computer

Northflank, a cloud application hosting platform, suffered a security breach resulting in the unauthorized decryption and theft of customers' environment variables, including AI API keys. The company warned all customers to immediately rotate stored credentials following a threat alert from Wiz. Multiple customers reported abuse of their AI API keys, with attackers potentially leveraging the stolen credentials for 'LLMJacking'—exploiting access to run up costly AI service usage. The breach is part of a broader wave of attacks on AI development platforms observed in October 2026, collectively dubbed 'Black October' by Wiz.

1w ago · security-com

Attackers have developed an autonomous offensive agent named 'Skynet' that hijacks exposed large language model (LLM) endpoints to conduct self-directed cyberattacks. The agent leverages stolen or unsecured AI model access to perform internet-wide scanning, exploit web frameworks, steal credentials and databases, and deploy cryptocurrency miners without human intervention. It uses sophisticated jailbreak techniques to bypass commercial model safety controls by redefining malicious activities as routine development tasks. The activity represents a novel form of LLMjacking, where attackers shift computational costs to victims while maintaining anonymity and evading provider bans.

1w ago · security-com

Two critical use-after-free vulnerabilities in PHP, tracked as CVE-2026-109618 and CVE-2026-109619 and collectively named PHP2Shell, enable unauthenticated remote code execution on affected servers. The flaws exist in the PHP runtime and can be exploited via crafted web requests, allowing attackers to take control of servers, execute arbitrary code, read sensitive memory, or establish persistence. Public proof-of-concept code emerged before patches were available, and exploitation attempts have been observed since October 17, 2026, with increased activity following disclosure. All PHP installations handling web requests are affected, especially those using Apache with mod_php, and immediate patching is strongly advised.

1w ago · securelist

Two critical remote code execution (RCE) vulnerabilities, CVE-2026-109618 and CVE-2026-109619, were disclosed in PHP, affecting versions 5 through 8. The flaws, related to use-after-free and memory corruption in the handling of multipart/form-data with session.upload_progress, allow unauthenticated attackers to execute arbitrary code on vulnerable web servers. An exploit named PHP2Shell has been demonstrated in the wild against popular PHP applications including WordPress, MediaWiki, and Drupal. Exploitation is highly likely immediately after patch release, and mitigation requires immediate patching or temporary configuration changes such as disabling upload progress and blocking multipart requests.

1w ago · securelist

In early 2026, the APT41 threat actor revived techniques from the BlackLotus UEFI bootkit to target diplomatic organizations in Central Asia, specifically in Tajikistan, Uzbekistan, and Turkmenistan. The attackers deployed a new bootkit dubbed BlackX, exploiting the Baton Drop vulnerability (CVE-2022-21894) to bypass Secure Boot by installing custom Machine Owner Keys (MOK). The attack involved a multi-stage payload, including a malicious kernel driver signed with a revoked certificate and user-mode implants such as a cloud-based RAT and a Rust-based C2 client communicating with www.archivetj.com.