Intelligence · Updated daily
AI-analysed threats, vulnerabilities and campaigns. Not just what happened — what it means, who's affected, and what to do about it.
An unauthenticated attacker can inject malicious JavaScript into WebSocket broadcast messages that execute in the browsers of all connected administrators. The vulnerability stems from unsanitized query parameters being stored and propagated to authenticated users without validation.
A compromised npm package (@cap-js/openapi v1.4.1) was published containing malicious code that harvests machine-local credentials and attempts self-propagation. Organizations must assume all credentials accessible during installation are compromised.
Unsanitized user input in the log_file_name parameter allows command injection into Stata execution contexts, enabling arbitrary code execution. This PoC demonstrates a critical trust boundary violation in parameter handling that bypasses existing security guards.
Cisco Talos published a conference attendance report from Cisco Live U.S. focusing on networking opportunities, AI discussions, and attendee wellness rather than security research findings or threat intelligence.
Cisco disclosed CVE-2026-20245, a root-level remote code execution vulnerability in SD-WAN products currently exploited in the wild with no patch available. This marks the seventh SD-WAN zero-day from Cisco in 2026, indicating either a sustained adversarial campaign or significant product security deficiencies.
The PCPJack threat actor has compromised approximately 230 servers across AWS, Google Cloud, and Azure to establish a covert SMTP relay network for sending spoofed or spam emails. The campaign reflects systemic credential theft and weak access controls across major cloud providers.
Shyam Sankar, CTO of Palantir Technologies, is a leading candidate to head the US Cybersecurity and Infrastructure Security Agency. This reflects potential policy shifts toward closer collaboration between federal cybersecurity infrastructure and commercial data analytics platforms.
Brave Software has released Origin, a paid subscription browser that removes cryptocurrency rewards, AI features, and other monetisation mechanisms from the standard Brave browser. This reflects growing user frustration with feature bloat and signals a viable commercial model for stripped-down privacy tools.
ESET researchers have documented direct operational cooperation between Gamaredon and Turla, two FSB-linked threat actors, with Gamaredon facilitating initial access for Turla against Ukrainian targets in 2025. This represents an unusual departure from typical competitive dynamics between state-sponsored groups and suggests coordinated Russian intelligence operations.
The U.S. executive order establishes a federal vetting process allowing up to one month pre-release review of advanced AI models for national security risks before public deployment. This represents a significant shift toward regulatory oversight of frontier AI systems.
Google patched 124 Android vulnerabilities in June 2026, including CVE-2025-48595, a high-severity privilege escalation flaw in the Framework component already under active exploitation in targeted attacks. This represents an unauthenticated, interaction-free privilege escalation affecting a core system component.
The White House has issued a pared-back executive order on AI that establishes confidentiality, cybersecurity, and intellectual property safeguards for federal access to AI models. This represents a policy shift toward managed risk rather than comprehensive regulation, with direct implications for how government agencies will interact with commercial and internal AI systems.
Microsoft released Coreutils for Windows at Build 2026, bringing native Linux command-line utilities to Windows. This development simplifies cross-platform development workflows but introduces new attack surface and dependency management concerns for Windows environments.
Vitest UI server on Windows fails to properly normalize file paths before security checks, allowing unauthenticated attackers to read arbitrary files via the `/__vitest_attachment__` endpoint when the server is network-exposed. This affects developers running Vitest in UI or Browser Mode on Windows systems.
PraisonAI Platform allows any workspace member to inject arbitrary users (including attacker-controlled accounts) as workspace owners by bypassing role-assignment authorization checks. This enables immediate privilege escalation and cross-tenant compromise within a workspace.
Attackers discovered a method to manipulate Meta's AI support assistant into resetting Instagram account passwords without proper verification, leading to defacement of prominent accounts including those of the Obama White House and U.S. Space Force. Instructions circulated on Telegram, indicating active exploitation in the wild.
CVE-2026-8732 in WP Maps Pro allows unauthenticated attackers to create administrative accounts, granting complete control of affected WordPress sites. Active exploitation has been observed in the wild.
Dashlane disclosed a brute-force attack in May 2026 where threat actors bypassed 2FA on fewer than 20 personal subscription accounts and downloaded encrypted vaults. The incident highlights authentication vulnerabilities in password managers and raises questions about 2FA implementation robustness.
NIST's National Vulnerability Database has fallen critically behind in processing new vulnerabilities, with unprocessed items doubling from 13,000 to 27,000 between February 2024 and end of 2025, according to an inspector general report. This operational failure directly undermines the NVD's utility as the primary source of truth for vulnerability data across the security industry.
Threat actor DriveSurge has compromised thousands of websites to distribute ClickFix and FakeUpdate malware variants. The scale and persistence of this campaign indicates sophisticated compromised-site infrastructure being repurposed for high-volume malware distribution.
YARA-X, the modernised Rust implementation of the YARA malware detection language, reached version 1.17.0 with performance improvements and a single bugfix. This routine release reflects ongoing maintenance of a critical tool in the security analyst's arsenal.
A previously unidentified threat actor is distributing NetSupport RAT, a legitimate remote access tool repurposed for malicious purposes. This represents an active malware campaign using RAT tooling for post-compromise control.
Dutch law enforcement and the NCSC successfully dismantled a botnet commanding at least 17 million infected devices across multiple platforms, with over 200 command-and-control servers operating from the Netherlands. This represents a significant disruption to a large-scale criminal infrastructure, though the source and purpose of the botnet remain unclear from available details.
A authentication bypass vulnerability in WP Maps Pro allows unauthenticated attackers to create administrative accounts on WordPress sites. Active exploitation is reported in the wild.
YAMCS yamcs-core 5.12.7 contains a user enumeration vulnerability allowing attackers to discover valid usernames through differential authentication responses. This information disclosure facilitates targeted account compromise and social engineering attacks.