We’re Tracking the Future of Security

Category: Cyber-Physical Security

Coverage of the convergence between IT/cybersecurity and physical security systems, including networked device risk, secure system integration and unified threat response.

  • WatchGuard Patches Five Critical Fireware OS Flaws That Could Let Attackers Seize Firebox Appliances

    WatchGuard Patches Five Critical Fireware OS Flaws That Could Let Attackers Seize Firebox Appliances

    WatchGuard has patched five critical vulnerabilities in its Fireware OS and WatchGuard Dimension software, including a set of unauthenticated flaws in the VPN negotiation process that could allow an attacker to remotely execute code on affected Firebox security appliances, according to security advisories the company updated through September 2026.

    The Vulnerabilities

    Three of the critical flaws — tracked as CVE-2026-19313, CVE-2026-19315 and CVE-2026-19318, each carrying a CVSS score of 9.3 — affect the iked process that handles Internet Key Exchange (IKEv1 and IKEv2) for IPsec VPN negotiation. The issues include a heap buffer overflow, a type confusion flaw and a stack-based buffer overflow. Because the affected process is exposed to unauthenticated network traffic, WatchGuard warned that a remote attacker could trigger the defects with specially crafted packets, without needing valid credentials, potentially leading to code execution on the device.

    Two further critical vulnerabilities affect WatchGuard Dimension, the company’s network visibility and reporting platform, and could allow an attacker to take over an administrator account.

    Fixed Versions

    WatchGuard has fixed the flaws in Fireware OS versions 2026.2.2, 12.12.2 and 12.5.20, and in WatchGuard Dimension version 2.3.1. The company said it has not observed the vulnerabilities being exploited in the wild, but urged administrators managing Firebox appliances to update immediately given the unauthenticated, pre-patch attack surface.

    Why It Matters

    Firebox appliances sit at the network edge for a large number of small and midsize organizations, making unauthenticated remote-code-execution flaws in VPN-handling code especially attractive to opportunistic attackers scanning the internet for exposed devices. Network security appliances of this kind have previously appeared on federal exploited-vulnerability tracking lists tied to broader intrusion campaigns, underscoring why perimeter security appliances are routinely targeted once technical details of a flaw begin to circulate. Security teams operating Firebox devices should prioritize patching over other maintenance work and review VPN gateway logs for anomalous IKE negotiation traffic predating the update.

  • Foreign Hackers Breach Two Colorado Water Utilities, Alter Pump Cycles and Disable Alarms

    Foreign Hackers Breach Two Colorado Water Utilities, Alter Pump Cycles and Disable Alarms

    Foreign hackers breached the computer systems of two small Colorado water utilities last month, changing pumping cycles, disabling alarms and altering equipment settings before operators regained control, state officials disclosed on September 19, 2026.

    What Happened

    According to Colorado officials, the intrusions did not compromise drinking water quality or treatment processes, and the affected utilities’ operators were able to identify and reverse the unauthorized changes. Governor Jared Polis’s office described the incidents as “brief” and said the associated risks were “quickly addressed by the providers themselves.” The identities of the specific utilities have not been publicly disclosed.

    Officials have not attributed the intrusions to a specific threat actor, though the state noted that previous, similarly styled attacks on US water infrastructure this year have been linked to actors with suspected ties to Iran and China.

    Part of a Wider Pattern

    The Colorado incidents add to a year in which more than 100 drinking water and wastewater systems across at least a dozen US states — including Minnesota, Michigan, Georgia, South Dakota and New Jersey — have reported cyberattacks targeting internet-exposed programmable logic controllers (PLCs). In late July, hackers broke into the operational technology behind more than 30 municipal water systems in Minnesota, changing passwords and IP addresses on exposed PLCs and locking operators out of their own equipment.

    The pattern echoes an earlier wave documented in coordinated attacks that disabled a UK power plant and hit US water utilities, and follows a related warning about active reconnaissance against internet-exposed Siemens S7 PLCs flagged by CISA, the NSA and the FBI.

    Federal Response

    The US Environmental Protection Agency (EPA) said it is investigating the Colorado breaches and working with the affected utilities to identify vulnerabilities and strengthen defenses. The agency has previously reported identifying more than 900 vulnerabilities across over 650 water systems nationwide in 2026, and says it has helped remediate roughly 700 of those weaknesses across more than 500 utilities.

    Why It Matters

    Water and wastewater utilities are frequently smaller, resource-constrained operators with internet-exposed industrial control systems and limited dedicated cybersecurity staff, making them attractive, comparatively low-effort targets for state-linked and opportunistic threat actors alike. Security researchers and federal advisories have repeatedly urged utilities to take PLCs off the public internet, enforce multifactor authentication for remote access, and monitor for unauthorized changes to ladder logic and equipment configurations.

  • Critical Docker Sandboxes Flaw Let Malicious Code Escape Into macOS Host Files

    Critical Docker Sandboxes Flaw Let Malicious Code Escape Into macOS Host Files

    Docker disclosed and patched a critical vulnerability, CVE-2026-77179, affecting Docker Sandboxes versions 0.28.0 through 0.41.x on macOS. The flaw sits in the virtio-fs host server that Docker Sandboxes uses to share files between a virtual machine and the host: when the server reopens an unlinked file from a previously stored path, it improperly follows symbolic links. A malicious guest can exploit that behavior by replacing a parent directory with a symlink, breaking out of the shared workspace it was supposed to be confined to.

    Once outside the shared workspace, code running inside the sandboxed VM can read or modify arbitrary files anywhere on the host, running with the privileges of the host account that operates the virtual machine, which can extend to host code execution. Docker fixed the issue in version 0.42.0, released September 7, 2026. Docker has not reported any in-the-wild exploitation, and as of the September 16 edition of CISA’s Known Exploited Vulnerabilities catalog, the flaw had not been added.

    Why it matters

    Sandboxing is supposed to be the containment boundary that makes it safe to run untrusted or AI-agent-generated code; a host-escape flaw in the sandbox’s own file-sharing layer undermines that boundary for any developer or CI pipeline that assumed the VM could not touch files outside its assigned workspace.

    Sources

  • Plugin4Shell: Same Zero-Click RCE Flaw Hits Claude Code, Codex, Copilot and Gemini CLI

    Plugin4Shell: Same Zero-Click RCE Flaw Hits Claude Code, Codex, Copilot and Gemini CLI

    Security researchers disclosed a zero-click remote-code-execution flaw, dubbed Plugin4Shell, affecting the plugin-installation logic shared by four major AI coding agents: Claude Code, OpenAI Codex, GitHub Copilot and Gemini CLI. Researchers describe it as the first supply-chain vulnerability found across the AI coding agent ecosystem, giving an attacker who exploits it the same system and data access as the developer running the agent.

    The flaw lies in how these tools handle SHA-pinned plugin versions: marketplaces pin a plugin to a specific Git commit hash after review, but for Claude Code, Codex and Copilot, an attacker can create a branch whose name matches that 40-character commit hash, and Git may check out the attacker-controlled branch instead of the reviewed commit. Gemini CLI has a related variant that abuses a branch named FETCH_HEAD to redirect the checkout. Anthropic patched Claude Code in version 2.1.179 and OpenAI fixed Codex in version 0.146.0. Microsoft has not yet shipped a fix for Copilot, and Google chose to deprecate Gemini CLI rather than patch it, leaving existing installs exposed indefinitely.

    Why it matters

    AI coding agents now run with broad repository and system access inside many development pipelines; a shared flaw in how four of the most widely used agents install plugins means a single exploitation technique can reach into a large share of the software supply chain that depends on them.

    Sources

  • Oracle Ships Java 27 With Post-Quantum Hybrid Key Exchange for TLS 1.3

    Oracle Ships Java 27 With Post-Quantum Hybrid Key Exchange for TLS 1.3

    Oracle released Java 27, the platform’s second non-LTS update since JDK 25, headlined by JEP 527: Post-Quantum Hybrid Key Exchange for TLS 1.3. The feature implements the IETF’s Hybrid Key Exchange in TLS 1.3 specification, pairing a quantum-resistant key-exchange algorithm with a conventional one in the same handshake so a connection stays secure as long as either algorithm holds up. The release ships nine JEPs in total, five of which remain in preview or incubator status.

    The hybrid approach is designed to blunt “harvest now, decrypt later” attacks, in which adversaries capture encrypted traffic today and store it until a sufficiently powerful quantum computer can break the classical encryption protecting it. Because Java 27 combines the new algorithm with an existing one rather than replacing it outright, Oracle says organizations can adopt the protection with minimal disruption to existing TLS-dependent applications.

    Why it matters

    TLS underpins the great majority of encrypted communication between servers, applications and connected devices; building quantum-resistant key exchange into a mainstream runtime like Java gives large swaths of enterprise and industrial software a path to post-quantum protection without a wholesale cryptographic migration.

    Sources

  • CISA Advisory Flags Two Critical Flaws in Hitachi Energy Grid Control Platform

    CISA Advisory Flags Two Critical Flaws in Hitachi Energy Grid Control Platform

    CISA published an advisory covering five vulnerabilities in Hitachi Energy’s FACTS Control Platform (FCP), software used to manage Flexible AC Transmission Systems that stabilize voltage and power flow on electric grids. Two of the flaws, CVE-2024-4872 and CVE-2024-3980, are rated 9.9 (Critical) on the CVSS v3.1 scale; a third, CVE-2024-7940, is rated 8.3 (High); a fourth, CVE-2024-3982, is rated 8.2 (High); and a fifth, CVE-2024-7941, is rated 4.3 (Medium).

    According to CISA, the flaws include improper neutralization of special elements in data query logic, path traversal, authentication bypass by capture-replay, missing authentication for a critical function, and open redirect. An authenticated attacker could chain these to inject code, access critical files, hijack sessions, bypass authentication or run phishing attacks against operators. Hitachi Energy says only FCP deployments with the GWS component installed since 2020 are affected, spanning FCP versions 3.4.0 through 4.1.1. The company’s advisory 8DBD000229 outlines remediation; interim mitigations include isolating control systems behind firewalls, requiring VPNs for remote access and enforcing strong password policies.

    Why it matters

    FACTS platforms sit inside the control layer that keeps transmission voltage stable — CISA lists the affected sector as Energy, and a successful chained exploit could let an authenticated attacker manipulate grid-stabilization logic rather than simply steal data.

    Sources

  • CISA Flags Command-Injection and SQL Injection Flaws in Schneider Electric NetBotz Monitors

    CISA Flags Command-Injection and SQL Injection Flaws in Schneider Electric NetBotz Monitors

    CISA published an advisory covering two vulnerabilities in Schneider Electric’s NetBotz 5 appliances, which provide video, door-contact, leak, smoke, vibration, temperature and humidity monitoring for data centers and industrial facilities. CVE-2026-13336 is an OS command-injection flaw reachable through a maliciously modified backup-restore file, rated 6.4 (Medium) on the CVSS v3.1 scale. CVE-2026-13337 is a SQL injection flaw in the appliance’s web interface, rated 4.6 (Medium).

    Both vulnerabilities affect NetBotz software versions 5.5.2 and earlier. Schneider Electric has released version 5.6.0 to remediate both issues. CISA lists the affected sectors as Commercial Facilities, Critical Manufacturing and Information Technology, with unpatched devices at risk of arbitrary code execution, device manipulation or unauthorized data access.

    Why it matters

    NetBotz appliances are themselves security and environmental monitoring equipment installed inside data centers — a vulnerability in the monitor undermines the very physical-security function the device is deployed to provide.

    Sources

  • Coast Guard and FBI Board Oil Tankers After Cyberattacks Reach Engine-Room Systems

    Coast Guard and FBI Board Oil Tankers After Cyberattacks Reach Engine-Room Systems

    The US Coast Guard and FBI boarded at least two foreign-flagged, US-bound oil tankers in the Gulf of Mexico — including the Liberian-flagged VL Prosperity — on August 21 and August 24, 2026, after detecting evidence that the vessels’ onboard networks had been compromised. Coast Guard Rear Adm. Amy Grable confirmed “evidence of a malicious cyber actor” on the ship’s IT/OT systems, though the Coast Guard has not attributed the intrusion to any state actor. A joint team of Coast Guard cyber specialists, a vessel inspector and an FBI Cyber Action Team spent four days aboard the VL Prosperity.

    Iran’s Mehr News Agency separately reported that the VL Prosperity suffered a cyberattack on August 7 while transiting the Strait of Gibraltar, alleging the intrusion reached engine-room systems — altering coolant flow, engine speed and fuel delivery — and knocked out communications for roughly 30 hours; those specific technical claims have not been confirmed by the Coast Guard. Officials say there is no evidence the vessels operated unsafely, and the Coast Guard says it has conducted an estimated 40-50 similar cyber-related boardings over the past year.

    Why it matters

    Commercial vessels increasingly run networked engine, navigation and fuel-management systems that were not designed with the same threat model as IT networks — an attacker reaching those systems mid-voyage, even briefly, raises the kind of physical-safety stakes that a typical office network breach does not.

    Sources

  • Ransomware Attacks on Manufacturers Rose Nearly 40% in First Half of 2026, Report Finds

    Ransomware Attacks on Manufacturers Rose Nearly 40% in First Half of 2026, Report Finds

    Cyber-risk intelligence firm Black Kite reported 1,183 manufacturing-sector ransomware victims in the first seven months of 2026, a nearly 40% increase year-over-year, with attacks on the sector more than doubling since 2023. Manufacturing accounted for 22% of all 7,551 ransomware disclosures the firm tracked across all industries in the period.

    The report found newer ransomware groups — including The Gentlemen, which alone claimed 142 manufacturing victims, along with Qilin, Akira, DragonForce and INC Ransom — have displaced the threat-actor groups that previously dominated the sector. Black Kite said 73% of attacks hit mid-market companies, with a median victim revenue of $42.9 million, and that European victim counts rose 85.4% year-over-year while the US share of global victims fell from 52.3% to 34.8%.

    Why it matters

    The shift toward mid-market manufacturers suggests attackers are increasingly targeting companies that may have less mature security operations than large enterprises but still enough revenue — and enough dependence on continuous production — to make ransom payment attractive.

    Sources

  • Hardcoded Cryptographic Keys in Wärtsilä Ship Management Software Rated Critical

    Hardcoded Cryptographic Keys in Wärtsilä Ship Management Software Rated Critical

    CISA disclosed two critical vulnerabilities in Wärtsilä FOS-Onboard, fleet-management software used aboard vessels, after the flaws were reported by maritime cybersecurity firm Cydome Security. CVE-2026-78225, rated 9.1 to 9.5 depending on scoring version, is a hardcoded cryptographic server key in the software’s Update Controller component that could let an attacker deliver an unauthorized software update or execute code on affected systems.

    A second flaw, CVE-2026-81855, is a hardcoded client authentication key in a robot-testing-framework component that could let an attacker extract credentials and impersonate a privileged client. Wärtsilä said the vulnerabilities are not exploitable under its recommended installation configuration and has made a patch available to customers on request. CISA lists the affected sector as Transportation Systems, with deployment worldwide.

    Why it matters

    Hardcoded keys are a recurring failure mode in operational technology because they can’t be rotated the way a normal credential can — once exposed, every deployed unit sharing that key is affected until firmware is patched.

    Sources