Hide Your RDP: Password Spray Leads to RansomHub Deployment

DFIR Report traces a full RansomHub kill-chain from initial RDP password-spray to domain-wide encryption, offering defenders detection timestamps and IOCs at every ATT&CK stage.

Version 1.1: Cisco Firewalls mit persistenter Malware infiziert

BSI updated its advisory on Cisco firewalls compromised with persistent malware, indicating the campaign scope or remediation guidance has expanded since the initial release.

FortiBleed campaign used custom FortiGate sniffer to steal credentials

The FortiBleed campaign deployed custom packet sniffers on compromised FortiGate firewalls to harvest authentication credentials in transit — elevating the urgency for any organisation still running unpatched Fortinet edge devices.

Cat’s Got Your Files: Lynx Ransomware

DFIR Report details a Lynx ransomware intrusion originating from a single exposed RDP session—no brute-force needed—underscoring the danger of any internet-facing remote-access without MFA.

vSphere and BRICKSTORM Malware: A Defender's Guide

Google's defender guide for BRICKSTORM malware targeting VMware vSphere environments provides actionable detection and hardening advice — critical for enterprises relying on vCenter as their virtualisation backbone.

Version 1.1: SonicWall Gen 7 Firewalls - Ransomware-Angriffe beobachtet

BSI confirms ransomware attacks targeting SonicWall Gen 7 firewalls, reinforcing the trend of threat actors systematically compromising perimeter appliances as their preferred initial-access method.

Investigating an adversary-in-the-middle phishing campaign targeting Microsoft 365 and Okta users

Datadog details an adversary-in-the-middle phishing campaign stealing M365 and Okta session tokens — a direct threat to enterprises relying on federated identity without phishing-resistant MFA.

Web Traffic Hijacking: When Your Nginx Configuration Turns Malicious

Datadog documents an active campaign hijacking web traffic via tampered NGINX configs and compromised Baota management panels — with IOCs to check against your own reverse-proxy infrastructure.

Risky Business #830 -- LiteLLM and security scanner supply chains compromised

Supply-chain compromises of LiteLLM and security-scanning toolchains show attackers increasingly targeting developer infrastructure, with one campaign bundling an anti-Iran wiper alongside the backdoor.

Risky Business #784 -- GitHub supply chain attack steals secrets from 23k projects

A cascading GitHub Actions supply-chain attack exfiltrated secrets from 23,000 repositories, demonstrating how a single compromised CI/CD component can propagate at ecosystem scale.

Microsoft links Mastra AI supply chain attack to North Korean hackers

Microsoft attributes the Mastra AI npm supply-chain compromise—140+ malicious packages—to North Korea's Sapphire Sleet, signalling that state-sponsored actors are now systematically targeting the AI-tooling ecosystem.

Risky Business #788 -- Trump targets Chris Krebs, SentinelOne

The Trump administration's political targeting of former CISA director Krebs and SentinelOne raises concerns about the independence of US cybersecurity institutions that European partners rely on for threat intelligence.

Russian Attackers Weaponize WinRAR Flaw Against Ukrainian Orgs

Two Russian APT campaigns are exploiting CVE-2025-8088 in WinRAR — patched a year ago — against Ukrainian military and government targets, underscoring the long tail of unpatched archive-handler vulnerabilities in conflict zones.

Chinese Hackers Abused Google Workspace Rules to Steal Research and Defense Emails

A Chinese espionage group abused Google Workspace mail-routing rules to silently exfiltrate defence and medical research emails for over a year after backdooring REDCap servers.

[UPDATE] [mittel] MISP: Mehrere Schwachstellen

MISP threat intelligence platform vulnerabilities could compromise security teams' ability to share and analyze threat indicators effectively.

Iran Signed a Ceasefire — Its Hackers Didn't

Iranian state-linked cyber operations persisted despite a formal ceasefire, spotlighting the gap in international law where cyber-hostilities are not explicitly covered by ceasefire obligations.

Phishing Attack Volume Down 20%, But Risk Still Rising

Phishing volumes dropped 20%, yet per-campaign success rates are climbing as attackers leverage AI to craft more convincing lures — a quality-over-quantity shift that defeats volume-based detection heuristics.

Stopping Sobolan Malware with Aqua Runtime Protection

A multi-stage campaign dubbed Sobolan targets Jupyter Notebook environments with compressed payloads from remote C2 servers, underscoring the growing risk to data-science infrastructure.

Tomcat in the Crosshairs: New Research Reveals Ongoing Attacks

Active exploitation campaigns against Apache Tomcat are accelerating, with attackers weaponising new vulnerabilities within hours—organisations should verify Tomcat exposure and patch status now.

Pass the Cookie and Pivot to the Clouds

Stolen browser session cookies let attackers bypass MFA and pivot from compromised endpoints into cloud services — a technique enterprises must detect at the session layer, not just the perimeter.

Agentjacking Attack Tricks AI Coding Agents Into Running Malicious Code

Social engineering attacks against AI coding assistants could compromise developer workstations by exploiting the trust relationship between humans and AI tools.

Copilot 'SearchLeak' Attack Allows 1-Click Data Theft

A now-patched three-stage prompt-injection chain in Microsoft Copilot used hidden URLs to exfiltrate user data with a single click—illustrating systemic risks in enterprise AI assistant deployments.

What we learned about TEE security from auditing WhatsApp's Private Inference

Trail of Bits' audit of Meta's TEE-based AI inference reveals hard lessons for enterprises seeking to combine end-to-end encryption with cloud-hosted LLM processing.

US Gov asks Anthropic to ban 'foreign national' access to Fable, Mythos

The US government's emergency intervention against Anthropic's AI models marks the first known instance of regulatory authorities forcing immediate AI capability restrictions based on national security concerns.

Google Antigravity exfiltrates data via indirect prompt injection attack

Autonomous cars, drones cheerfully obey prompt injection by road sign

SkillMutator: Benchmarking and Defending Language-and-Code Cross-modal Attacks on LLM Agent Skills

New research demonstrates cross-modal attacks where adversaries manipulate both documentation and code to compromise LLM agent skills, creating enterprise risks for AI-powered automation workflows.

One-Click Microsoft 365 Copilot Flaw Could Have Let Attackers Steal Emails, Files, and MFA Codes

Varonis chained three bugs in Microsoft 365 Copilot Enterprise Search into a one-click exfiltration path — emails, files, and MFA codes were reachable via a trusted Microsoft URL.

Entra Agent ID: Inside a cross-tenant agent compromise

A cross-tenant agent compromise via Entra Agent ID blueprints mirrors the Midnight Blizzard attack pattern — an attacker controlling a third-party blueprint can impersonate any downstream agent, threatening multi-tenant SaaS environments.

Promoting Advanced Artificial Intelligence Innovation and Security

US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

Microsoft 365 Copilot Generated Images Accessible Without Authentication -- Fixed!

Microsoft 365 Copilot-generated images were publicly accessible without authentication until a fix was applied, underscoring that AI features bolted onto SaaS suites can quietly erode access controls.

The Normalization of Deviance in AI

Drawing parallels to the Challenger disaster's cultural failures, this essay argues the AI industry is normalising known security defects—a governance-level warning for boards green-lighting large-scale AI rollouts.

Show HN: Claw Patrol, a security firewall for agents

Dynamic Free-Rider Detection in Federated Learning via Simulated Attack Patterns

A new method uses simulated attack patterns to detect free-rider clients in federated learning, addressing a model-integrity gap relevant to enterprises deploying privacy-preserving ML under EU AI Act obligations.

CISA Warns of Actively Exploited Joomla JCE Flaw Allowing PHP Code Execution

CISA added a max-severity Joomla JCE flaw enabling unauthenticated PHP execution to its KEV catalogue, signalling active exploitation that affects the many EU SMBs still running Joomla-based sites.

Version 1.0: Ivanti Connect Secure - Zero-Day Angriffe beobachtet

BSI confirms active zero-day exploitation of Ivanti Connect Secure appliances, continuing a pattern of edge-device targeting that has plagued European organisations since late 2023.

Version 1.0: Kubernetes - Kritische Schwachstelle im Ingress NGINX Controller ermöglicht Clusterübernahme

BSI flags a critical Kubernetes Ingress NGINX Controller vulnerability that allows full cluster takeover — a severe risk for any enterprise running containerised workloads with exposed ingress endpoints.

Twenty One Zero-Days in FFmpeg

Fragnesia: Linux Kernel Local Privilege Escalation via ESP-in-TCP

Wiz discloses 'Fragnesia,' a new Linux kernel page-cache corruption bug in the ESP-in-TCP path enabling unprivileged local-to-root escalation — part of the growing Dirty Frag exploit family.

Persistent XSS/RCE using WebSockets in Storybook’s dev server

A WebSocket hijacking vulnerability in Storybook's development server (CVE-2026-27148) can be chained to achieve persistent XSS and RCE, creating a credible path to supply-chain compromise for any team running Storybook locally.

Version 1.1: Citrix NetScaler ADC & NetScaler Gateway - Kritische Sicherheitslücken geschlossen und teils aktiv ausgenutzt

BSI reports critical Citrix NetScaler ADC and Gateway vulnerabilities are patched but some are already being actively exploited — urgency is high for organisations that delayed remediation.

[UPDATE] [hoch] git: Mehrere Schwachstellen

Git vulnerabilities present supply chain risk through compromised version control, potentially affecting software integrity across development teams and CI/CD pipelines enterprise-wide.

[UPDATE] [hoch] Bouncy Castle BC-JAVA: Mehrere Schwachstellen

Cryptographic library vulnerabilities in Bouncy Castle threaten the security foundation of countless Java applications across enterprise environments, potentially undermining encryption assumptions.

ShinyHunters Exploits Oracle PeopleSoft Zero-Day (CVE-2026-35273) to Breach Universities

University targeting suggests the group prioritizes high-value research data and academic credentials over traditional financial targets.

[NEU] [hoch] IBM i: Schwachstelle ermöglicht Ausführen von beliebigem Programmcode mit Administratorrechten

IBM i administrative privilege escalation vulnerability threatens mainframe environments still prevalent in German manufacturing and financial sectors.

[UPDATE] [hoch] Google Chrome / Microsoft Edge: Mehrere Schwachstellen

Chrome and Edge browser vulnerabilities require immediate patching across enterprise endpoints to prevent code execution through web-based attacks.

[UPDATE] [hoch] Ubiquiti UniFi OS: Mehrere Schwachstellen

Ubiquiti UniFi network management flaws enable privilege escalation and data manipulation across enterprise wireless infrastructure deployments.

TikTok, AliExpress and WeChat ignore your GDPR rights

Delve – Fake Compliance as a Service – Part II

SBOMs in 2026: Everyone's generating them, no one's using them

ENISA's 2026 study of 334 organisations reveals that SBOM generation has become routine but operational consumption — vulnerability correlation, procurement gating — lags far behind, weakening CRA readiness.

Data sharing and GDPR compliance: Bounty UK shows what not to do

The UK ICO's enforcement against Bounty for illegal bulk data sharing offers a concrete precedent for EU-operating firms to benchmark their own third-party data-sharing practices against GDPR Article 6 expectations.

Italy fines Eni Gas e Luce €11.5 million for multiple GDPR violations

Italy's €11.5M double fine against Eni Gas e Luce for consent and telemarketing violations remains one of the largest early GDPR penalties — a reference point for enterprises auditing consent management workflows.

I Read NIST 800-218 So You Don’t Have To: Here’s What To Watch Out For

Practical breakdown of NIST SP 800-218 requirements, highlighting the areas most likely to create compliance friction for organisations not yet following attestation-based development practices.

How to explain the CISA software attestation requirements to your board

CISA's draft self-attestation form translates OMB M-22-18 into concrete minimum requirements—a useful benchmark even for EU firms selling software to US federal buyers.

Building digital products for the Cyber Resilience Act

Chainguard positions its zero-CVE container approach as a shortcut to meeting the EU CRA's secure-by-design mandate — useful vendor framing, but teams should map actual CRA obligations independently.

Data anonymization and GDPR compliance: the case of Taxa 4×35

Danish taxi firm Taxa 4×35's enforcement case illustrates how DPAs scrutinise anonymisation techniques — a useful benchmark for organisations processing high-volume personal data under GDPR Art. 89.

What the first Italian GDPR fine reveals about data security liabilities for processors

Italy's €50k fine against the Rousseau platform underscores that data processors — not only controllers — bear direct GDPR liability when security controls are inadequate.

Spring 2026 SOC 1 and 2 reports are now available in OSCAL format

AWS now publishes SOC 1 and SOC 2 reports in machine-readable OSCAL format, enabling automated ingestion into GRC platforms and reducing the manual effort of continuous cloud-provider assurance reviews.

Supply Chain Security Risk: GitHub Action tj-actions/changed-files Compromised

CVE-2025-30066 in the popular tj-actions/changed-files GitHub Action leaked CI/CD secrets through build logs, echoing earlier supply-chain compromises and reinforcing the need for pinned action references.

Living Off the Pipeline: Defending Against CI/CD Subversion

Deep-dive into 'living-off-the-pipeline' techniques where attackers abuse legitimate CI/CD features—build triggers, artefact caches, runner tokens—rather than injecting malicious code directly.

[tl;dr sec] #321 - Sandboxing AI Agents, Trivy Compromised, Pentesting AWS' AI Pentester

The compromise of Trivy, a widely-deployed container scanner, underscores that even security tooling in CI/CD pipelines is a viable supply-chain target, while AWS's own AI-based pentesting agent showed exploitable weaknesses.

Novo Nordisk Breach Highlights Software Development Pipeline Risk

A leaked GitHub token at Novo Nordisk exposed development pipelines, underscoring that secrets management must be treated as an identity-governance discipline—not just a vault configuration task.

We hardened zizmor's GitHub Actions static analyzer

After the Trivy-action supply-chain attack on PyPI, Trail of Bits hardened zizmor to detect pull_request_target misconfigurations that enable secret exfiltration from GitHub Actions.

Mini Shai-Hulud Strikes Again: TanStack + more npm Packages Compromised

The Mini Shai-Hulud campaign has compromised TanStack and other high-value npm packages, injecting malicious code into widely used developer tooling—a direct supply-chain threat to any JavaScript-heavy enterprise stack.

GitHub to Disable npm Install Scripts by Default to Stop Supply Chain Attacks

GitHub's breaking change reflects growing industry recognition that convenience features in package managers have become major supply chain attack vectors.

Dealing with multiple SBOMs

Practical strategies for merging and correlating multiple SBOMs across a product's lifecycle — a growing pain as CRA and US EO mandates push SBOM adoption into real workflows.

The security costs of base image version loitering

Stale base images silently accumulate known CVEs; the article quantifies how version-pinning without regular refresh inflates container attack surface over time.

Curl will not accept vulnerability reports during July 2026

We beat Google’s zero-knowledge proof of quantum cryptanalysis

Trail of Bits claims a more efficient zero-knowledge proof challenging Google's timeline for quantum ECC key-breaking, intensifying the post-quantum migration urgency.

V8 Heap Archaeology: Finding Exploitation Artifacts in Chrome’s Memory

SpecterOps details memory-level artifacts left by V8 JavaScript exploitation primitives, offering forensic analysts and exploit researchers a detection methodology for Chrome full-chain attacks.

npx Confusion: Packages That Forgot to Claim Their Own Name

Researchers registered 128 unclaimed npm names referenced in official docs and logged 121k downloads in seven months — demonstrating that namespace confusion in package managers remains a systemic supply-chain risk.

Coruna: The Mysterious Journey of a Powerful iOS Exploit Kit

GTIG's disclosure of the Coruna iOS exploit kit — covering five exploits against iOS 13 through 17.2.1 — underscores the commercial exploit market's ability to stockpile and chain zero-days across multiple iOS generations.

The Proliferation of DarkSword: iOS Exploit Chain Adopted by Multiple Threat Actors

GTIG documents DarkSword, a full-chain iOS exploit leveraging multiple zero-days and now adopted by several threat actors since late 2025 — demonstrating how exploit proliferation accelerates once a chain enters the commercial market.

Oops, I Weaponized the Database: Abusing AI Features in SQL Server 2025

SQL Server 2025's new AI features—vector search and external model calls—open practical channels for data exfiltration and C2 transport entirely within the database engine, with published PoC code.

Scripting the disassembler: Local agentic reverse engineering through vbdec’s live COM object model

Cisco Talos demonstrates a local agentic reverse-engineering workflow where AI agents interact with vbdec's live COM interface—a practical pattern for augmenting malware analysis without cloud-dependent LLMs.

Accelerate security investigations with Kiro CLI

AWS's Kiro CLI leverages AI to streamline security investigations by auto-generating complex CLI queries across multiple services — reducing analyst friction during time-critical incident response in cloud environments.

Most CISOs Report Pressure to Bury Bad Security News

Surveys show most CISOs face implicit pressure to delay or downplay security disclosures—a governance gap that NIS2's 24-hour notification mandate will make legally untenable.

Maine breach portal abused to publish fake data breach disclosures

Fraudulent breach disclosures were posted on Maine's official breach portal before verification, forcing named companies into public denials — an emerging misinformation tactic that weaponises regulatory transparency processes.

Maine disables data breach notification portal after fake disclosures

Maine's breach notification system compromise exposes how public disclosure portals themselves become attack vectors for spreading disinformation and undermining regulatory transparency.

Reliable CVE sources in the age of NIST NVD cutbacks

With NIST stepping back from CVE enrichment, enterprises must diversify vulnerability intelligence sources or risk blind spots in patch-prioritisation workflows.

Holding blobs for ransom: Four methods for Azure Storage ransomware

Four distinct ransomware vectors targeting Azure Blob Storage — including customer-managed key abuse and versioning manipulation — reveal that cloud-native data stores are not immune to extortion tactics traditionally aimed at on-prem infrastructure.