Find what you’re looking for
Browse everything
6,838 results in Threat Reports
When AI Becomes the Attacker: How a $25 Campaign Breached 27 Companies in a Week
In September 2026, a Chinese-speaking threat actor deployed three agentic AI frameworks — Hermes, Strix, and Cairn — against 105 e-commerce targets over five days, compromising at least 27 organizations, including a Fortune 500 company. The autonomous system stole more than 600,000 payment card records, planted credit-card skimmers on five sites, and gained unauthorized access across the remainder, all at an average cost of $25.46 per completed target. Parallel research by Silverfort demonstrated that a publicly available Claude Opus agent could escalate initial access to full domain control in eight minutes for roughly $12, confirming that agentic AI has industrialized the entire attack chain. This campaign signals a structural shift in the threat landscape: autonomous AI eliminates the skilled-attacker bottleneck, collapsing marginal attack costs to near zero. With initial-access prices on dark markets already falling from $1,400 to $429, and ransomware activity up 45% in 2026 despite fewer ransom payments, the economics now decisively favor attackers — demanding that defenders rethink architecture, not just tooling.
Frozen Out: How Unpatched Fridge Firmware Burned 700 European Supermarkets
In October 2026, attackers exploited unpatched firmware vulnerabilities in commercial refrigeration controllers — specifically CVE-2025-41450, CVE-2025-41451, and CVE-2025-41452 in Danfoss AK-SM 800A, as well as 23 flaws in Dixell/Copeland XWEB Pro — to sabotage refrigeration systems across more than 700 European grocery stores in a single night. Threat actors gained remote access via third-party service providers managing legacy OT equipment, manipulated defrost cycles to disguise the attack as routine maintenance, and forced temperatures above 50°C, destroying perishable inventory and triggering minor fires. Intermarché alone reported tens of millions of euros in losses, with impacts spanning France, Spain, Portugal, Poland, Germany, and Belgium. Attackers accessed systems through unmonitored third-party OT vendor connections on internet-exposed controllers running end-of-life firmware dating to 2016. This incident signals a sharp escalation in ICS/OT sabotage targeting civilian food infrastructure. With thousands of Danfoss systems still internet-exposed and researchers having published exploit details just weeks before the attack, the gap between vulnerability disclosure and weaponization is closing dangerously fast — raising urgent questions about third-party OT access governance across critical supply chains.
SBI's 2026 PII Leak: How a Small Breach Feeds India's Multi-Million-Record Underground Data Market
In October 2026, the State Bank of India (SBI) disclosed a customer data breach after a threat actor operating under the alias 'Siriusim' published 3,897 customer records—along with a table of 375,000 rows and 2,722 PDF files—on BreachForums. The exposed PII included names, mobile numbers, email addresses, dates of birth, home addresses, KYC documentation, Central KYC numbers, video KYC verification data, loan application metadata, and employment declarations. SBI confirmed the incident via a formal notice to India's National Stock Exchange, while stressing that core banking operations remained unaffected. The breach is suspected to have originated from a peripheral data-handling or digital-services system rather than SBI's core banking platform. This incident underscores the compounding risk that even small PII leaks create within India's mature underground data economy, where threat actors cross-reference breach dumps with real-time lookup services to build high-fidelity identity profiles for phishing, vishing, loan fraud, and impersonation attacks—amplifying the harm far beyond the raw record count.
Storm-0940's AI-Powered Supply Chain Attack: How Agentic AI Breached 2,250 Microsoft 365 Tenants Through ConnectWise
In October 2026, Microsoft attributed a large-scale supply chain attack against ConnectWise Cloud Backup to Storm-0940, a China-linked threat actor. Exploiting compromised application credentials tied to ConnectWise's authorized Microsoft 365 integrations, Storm-0940 deployed an agentic AI penetration-testing framework called ARTEX to automate reconnaissance, password spraying, and OAuth application abuse at scale—targeting more than 2,250 downstream customer tenants across manufacturing, aerospace, energy, defense, and financial services. In several cases, attackers achieved privileged cloud persistence in under three hours, generating up to 2,900 sign-in attempts per minute. ConnectWise suspended its Cloud Backup, SaaS Security, and Cloud Management services in response. This incident marks a watershed moment in AI-enabled offensive operations. Storm-0940's use of ARTEX demonstrates that nation-state actors are now weaponizing agentic AI to exponentially accelerate attack timelines, compress defender response windows, and scale supply chain compromise across thousands of organizations simultaneously—elevating the urgency of zero trust, segmentation, and multicloud visibility controls.
PHP2Shell: Millions of Exploit Attempts Targeting Critical PHP RCE Flaws to Deploy Mirai's Potassium Botnet
In October 2026, threat actors launched millions of exploitation attempts against two critical PHP vulnerabilities—CVE-2026-109618 and CVE-2026-109619 (CVSS 9.2)—targeting unauthenticated remote code execution via crafted HTTP requests. Attackers attempted to deploy the Mirai-derived Potassium botnet, which uses ChaCha20-encrypted configuration, ENS-based C2 resolution, and supports 24 DDoS flood commands across ICMP, TCP, UDP, and OpenVPN protocols. Cloudflare observed hundreds of thousands of exploit attempts within two hours of public disclosure, escalating to millions by October 22. Over 34.4 million PHP services remained vulnerable as of October 23, prompting precautionary shutdowns of government, hospital, and school websites across Asia, Europe, and North America. PHP2Shell is drawing comparisons to Log4Shell due to its deep engine-level origin dating to PHP 5.4 (2012), meaning no application-layer fix suffices. With 56.6 million web properties running PHP and patches unavailable for end-of-life branches, organizations face a prolonged exposure window. The rise of ASLR-bypass techniques and automated exploit toolkits signals escalating attacker sophistication targeting cloud-hosted and shared-hosting PHP environments.
Node2Shell: How a Crafted Brotli Payload Delivers Remote Code Execution on Unpatched Node.js Servers
In October 2026, security researchers at Imperva publicly demonstrated Node2Shell, a proof-of-concept remote code execution exploit targeting CVE-2026-111497, a memory corruption flaw in the Brotli decoder bundled with Node.js 22.x, 24.x, and 26.x. By feeding crafted malformed Brotli-compressed data through HTTP request bodies or response paths—including via Express 5 middleware such as express.json()—attackers can trigger an out-of-bounds write, ultimately executing arbitrary attacker-supplied code on unpatched servers. Node.js released patched builds on October 21, 2026; no in-the-wild exploitation had been confirmed as of October 22, 2026. This incident highlights the compounding risk of trusted, bundled dependencies in widely deployed runtimes. As Node.js underpins millions of cloud-native APIs and microservices, a single decoder flaw becomes a systemic supply-chain-adjacent threat. The rise of middleware-enabled attack surfaces—where standard Express body parsers silently expand exposure—underscores the urgency of runtime patching and network-layer inspection for modern cloud environments.
When AI Agents Pay the Attacker: Inside the Critical x402 SDK Vulnerabilities
In October 2026, security firm V12 disclosed 25 vulnerabilities in the x402 open-source SDK — a Coinbase-created, Linux Foundation-governed protocol enabling AI agents to autonomously make cryptocurrency micropayments during API calls. Three flaws were rated critical and all were silently patched in August 18 releases without public security advisories. The most severe allowed a malicious API to request an AI agent's entire token wallet balance, exploiting the default client's lack of spending limits. Additional critical flaws enabled a complete Solana payment bypass via token approval spoofing and cross-site login signature relay attacks. No package manager warnings were issued, leaving developers unaware of the risk. This incident signals a maturing threat surface around agentic AI and autonomous payment systems. As AI agents increasingly operate with financial autonomy — signing transactions without human approval — supply-chain vulnerabilities in foundational SDKs become high-stakes attack vectors. The combination of silent patching, absent CVE assignments, and no package manager advisories reflects a systemic disclosure gap the industry must urgently address.
CVE-2026-112900: How a Crafted Filename Turns Python's Most Popular Image Library Into a Remote Code Execution Risk
In October 2026, security researchers at OX Security disclosed three critical vulnerabilities in Pillow, Python's most widely used image processing library, downloaded approximately 436 million times per month. The most severe flaw, CVE-2026-112900 (CVSS 9.2), allows remote code execution via crafted EPS filenames passed unsanitized to Ghostscript, enabling an attacker who controls a filename to execute arbitrary OS commands on the server. Two additional critical flaws affect JPEG (FPX format) and TIFF in-memory processing pipelines. All three were patched in Pillow 13.0.0 released October 16, 2026. No active exploitation has been confirmed, though a proof-of-concept exists for the TIFF flaw. These vulnerabilities are emblematic of a broader supply-chain risk pattern: a single transitive dependency embedded across web upload forms, ML pipelines, document processors, and container images can silently expose thousands of applications. As AI-assisted vulnerability research (demonstrated here by Anthropic's Claude) accelerates flaw discovery, organizations must treat open-source image libraries as high-priority attack surface in their software supply chain programs.
MythicRatatoskr & Yggdrasil: How Sandworm Used a Mesh Network Backdoor to Wipe Ukrainian Infrastructure
In July 2025, Russian-aligned threat actor UAC-0099 (aka Earth Sirrush) gained initial access to a Ukrainian logistics company via a spear-phishing email deploying the MATCHWOK backdoor. A day later, the group introduced a previously undocumented implant called MythicRatatoskr, which routed command-and-control traffic through the Yggdrasil peer-to-peer mesh network to evade detection. Over two months later, on September 17, 2025, UAC-0099 handed off access to Sandworm, which executed a destructive data-wiping attack using the SDelete utility pushed via Active Directory Group Policy—destroying operational data across the compromised environment. This incident marks the first documented abuse of the Yggdrasil mesh network by a state-aligned threat actor for C2 infrastructure, signaling a new evasion frontier. The coordinated handoff between UAC-0099 and Sandworm reflects a maturing division-of-labor model among Russian APT groups, raising the operational risk bar for critical infrastructure defenders globally.
70,000 Credentials Stolen: How Threat Actor k0t Industrialized Enterprise Firewall and VPN Exploitation in 2026
Between February and June 2026, a threat actor tracked as k0t (aliases: CryptoFeel, Member19, QazWsxShop) executed a large-scale credential harvesting campaign against internet-exposed Fortinet FortiGate and Palo Alto Networks GlobalProtect devices worldwide. Leveraging public scanning tools such as FOFA and ZoomEye, k0t enumerated vulnerable edge appliances and exploited CVE-2025-64446 (FortiWeb path traversal) and CVE-2026-0257 (PAN-OS authentication bypass) via bespoke Python scripts. Post-exploitation actions included Active Directory enumeration, persistence via MeshCentral, Atera, and AnyDesk, and data exfiltration through Rclone to Wasabi S3 and MEGA. Over 70,000 credentials were harvested, with victims spanning Asia Pacific, Latin America, Europe, and MENA. This campaign underscores the accelerating role of initial access brokers (IABs) feeding ransomware groups such as Qilin and The Gentlemen. With edge device exploitation now industrialized through automated scanning and off-the-shelf exploit scripts, organizations relying solely on perimeter firewalls face systemic credential exposure risk—making zero trust segmentation and encrypted east-west traffic controls critically urgent.
PHP's Double Vulnerability Crisis: What CVE-2026-115560 and PHP2Shell Mean for Your Cloud Infrastructure
In October 2026, PHP disclosed a second unpatched remote code execution vulnerability, CVE-2026-115560, just two days after emergency patches addressed the critical PHP2Shell flaws CVE-2026-109618 and CVE-2026-109619. The PHP2Shell vulnerabilities, patched on October 21 in versions 8.2.35 through 8.5.12, allowed unauthenticated attackers to execute code on web servers. Exploitation attempts surged from tens of thousands within 45 minutes of disclosure to millions by October 22, with threat actors attempting to deploy a Mirai botnet variant. CVE-2026-115560 affects a narrow configuration involving session serialization handler switching and requires disabling session upload progress as a temporary mitigation. This incident illustrates the compounding risk of staggered vulnerability disclosures in widely deployed open-source runtimes. The rapid weaponization of PHP2Shell flaws—with honeypot detections four days before official disclosure—signals increasingly shortened exploit development windows, forcing organizations to treat patch cadence and network-layer egress controls as non-negotiable security baselines.
PHP RCE Vulnerability 2024: What Executives Need to Know Before Mass Exploitation Begins
In October 2024, SANS Internet Storm Center elevated its Infocon alert level in response to a critical PHP vulnerability affecting a broad range of web-facing applications, embedded devices, and appliances. The flaw—tied to PHP's session.upload_progress mechanism—enabled remote code execution, prompting coordinated disclosure across vendors including Cloudflare and Amazon, both of which reported large volumes of exploit attempts in the wild. Although no fully weaponized public exploit was confirmed at time of publication, security researchers warned that working exploit code was imminent. Patches were issued by php.net and major vendors, but incomplete coverage of embedded and legacy devices left significant attack surface exposed. This incident highlights a growing pattern of critical web-runtime vulnerabilities being exploited before organizations can fully patch their environments. The coordinated disclosure process succeeded, yet unpatched embedded devices and delayed enterprise patch cycles continue to represent systemic risk—underscoring why network segmentation and east-west traffic controls are essential backstops when software updates cannot be applied immediately.