CYBER THREATCAST

Beyond the Check Point crisis, this period reflects a broader pattern of exploitation targeting network management planes and developer infrastructure. A critical unauthenticated RCE chain in Ubiquiti UniFi OS (CVE-2026-34908 through CVE-2026-34910) enables root-level compromise of network and physical security management systems via a single crafted HTTP request, while a high-severity Cisco Catalyst SD-WAN Manager flaw (CVE-2026-20245) is under active exploitation with potential links to previous UAT-8616 activity. SolarWinds Serv-U (CVE-2026-28318) has been added to CISA's KEV catalog following confirmed active exploitation of its unauthenticated denial-of-service condition, with a June 19 federal patch deadline in effect. The Linux kernel's nftables subsystem (CVE-2026-23111) is similarly under pressure, with public exploits now available for a use-after-free vulnerability enabling local privilege escalation to root on Ubuntu and Debian systems, and a critical Redis RCE flaw dubbed DarkReplica (CVE-2026-23631) threatens misconfigured deployments through Lua VM memory corruption triggered via the SLAVEOF replication mechanism.

The vulnerability disclosure volume across the broader ecosystem remains exceptionally high, with several developments warranting immediate operational attention. Google's Chrome 149 release addresses a record 429 security flaws including 22 critical use-after-free and memory safety defects, while Google's June 2026 Android security update patches 124 vulnerabilities including CVE-2025-48595, a Framework privilege escalation flaw confirmed under active limited exploitation. The Everest Forms Pro WordPress plugin (CVE-2026-3300, CVSS 9.8) is being actively exploited for unauthenticated PHP code execution, with over 29,000 documented exploit attempts observed since April. AI-accelerated vulnerability discovery is emerging as a structural force multiplier: autonomous agents have uncovered 21 zero-day vulnerabilities in FFmpeg and a critical four-year-old Zcash Orchard pool counterfeiting flaw was identified using Claude Opus, signaling that the window between vulnerability existence and discovery is compressing rapidly across both commercial and open-source software ecosystems.

State-nexus espionage activity remains broadly active across multiple vectors. China's military intelligence services are conducting systematic digital recruitment operations against journalists, academics, and policy experts via LinkedIn and professional networking platforms, with the Five Eyes alliance issuing warnings about the tactic's effectiveness in bypassing traditional counterintelligence defenses. VerdantBamboo's deployment of a BSD variant of BRICKSTORM malware on Linux appliances following an 18-month breach demonstrates the patience and persistence characteristic of state-sponsored intrusions, while a newly identified China-linked cluster designated OP-512 is targeting IIS servers with cryptographically unique web shell frameworks designed to evade existing detection signatures—with evidence of 75-day pre-deployment dwell time suggesting deliberate, long-cycle operational planning. An American citizen's guilty plea for acting as a paid Chinese MSS intelligence intermediary, compensated at least $100,000 to target the Department of Justice, provides a ground-level illustration of the human intelligence dimension accompanying technical cyber operations.

The supply chain and AI development ecosystem has emerged as a primary theater of threat actor activity, with the TeamPCP-originated Shai-Hulud campaign's Mini framework now operating independently via copycat actors who have weaponized validly-signed artifact production and install-time credential theft across both npm and PyPI ecosystems. The Hades campaign represents a further evolution targeting graph machine learning and bioinformatics packages with cross-platform memory scrapers, AI analyst misdirection capabilities, and a wiper deterrent—indicating sophisticated adversarial awareness of security triage methodologies. The new extortion group Pink (CL-CRI-1147), assessed as a potential BlackFile rebrand with tactical overlaps with Scattered Spider and Lapsus$, is targeting enterprise cloud storage credentials through vishing combined with real-time MFA bypass, establishing a dark web leak site on May 31 that signals intent for sustained operations. Collectively, these developments point to a threat environment where identity, developer trust, and AI integration pipelines have become the primary attack surface.

Agentic AI systems present a qualitatively different security challenge from conversational AI, as OWASP's 'State of Agentic AI Security and Governance v2.01' report and multiple independent analyses confirm. When AI agents possess tool access, API permissions, and real-world action capabilities, a single successful prompt injection can escalate from output manipulation to active system compromise spanning initial access through data exfiltration. Microsoft Threat Intelligence's identification of campaigns where threat actors impersonate ChatGPT, Copilot, DeepSeek, and Claude as social engineering lures—spanning phishing, malvertising, and SEO-driven attacks—illustrates how AI brand trust is being systematically weaponized. The Meta AI support chatbot incident, where an authorization and business logic failure allowed attackers to take over 20,225 Instagram accounts including high-profile targets by manipulating the AI into performing privileged account recovery actions without adequate verification, exemplifies OWASP's 'Excessive Agency' risk category in production deployment at scale.

AI-accelerated vulnerability discovery is rapidly reshaping the offensive security timeline, creating compressive pressure on defenders. Autonomous agents from security startup depthfirst uncovered 21 zero-day vulnerabilities in FFmpeg including heap and stack overflows dormant for up to 23 years, while Google's Chrome 149 release of 429 patches likely reflects AI-generated vulnerability report influence. Anthropic's Project Glasswing—which has reportedly identified approximately 23,000 potential vulnerabilities across 1,000 open-source projects using its Mythos model—has drawn significant criticism from Bruce Schneier and others for its opacity, with nearly none of the identified vulnerabilities patched and the company refusing to release detailed findings. The Picus Security verification under Anthropic's Cyber Verification Program, granting restricted Mythos-class access for autonomous penetration testing across reconnaissance, exploitation, lateral movement, and privilege escalation phases, and Reco's demonstration of an LLM agent autonomously discovering and exploiting high-severity vulnerabilities in Salesforce Experience Cloud sites, collectively signal that AI-enabled offensive capability is transitioning from research to production deployment across both security tooling and adversarial operations.

Beyond supply chain malware, several distinct campaigns reflect broader trends in financially motivated threat activity. IBM has identified OverlordMX, a new man-in-the-browser banking trojan campaign in Latin America that combines automated overlay injection with live operator monitoring, enabling real-time credential collection and remote device takeover before victims are aware of the attack. Operation FlutterBridge, attributed to cybercrime network CL-CRI-1089, distributes the FlutterShell macOS backdoor through fake Google advertisements, with the malware hijacking Chrome through Secure Preferences modification and featuring arbitrary command execution, file system interaction, and environment variable exfiltration across three observed versions since late 2025. The Android NFC relay malware ecosystem has surged 188% in early 2026, with Kaspersky blocking 35,600 attacks from malware leveraging Host Card Emulation to intercept payment terminal requests through attacker-controlled servers coordinated via Telegram bots.

Ransomware continues to inflict significant operational damage across diverse sectors, with educational institutions proving particularly vulnerable. Evanston Township High School suffered a ransomware attack requiring two-day closure and FBI cooperation, while the Evanston area has experienced multiple cyber incidents within a single month. The broader RaaS ecosystem reflects sustained professionalisation, with Qilin affiliates demonstrating the integration of sophisticated initial access capabilities—including zero-day VPN exploitation—with traditional ransomware deployment chains. The Payload and TheGentlemen ransomware groups claimed additional victims in the manufacturing and education sectors respectively, underscoring that the double extortion model continues to drive targeting decisions regardless of victim size or sector. A targeted infostealer campaign against Russian critical infrastructure using mail.ru Cloud distribution, a Rust-based dropper disguised as PDF, and Telegram API command-and-control illustrates concurrent state-adjacent offensive operations outside the RaaS model.

Adversary-in-the-Middle phishing platforms continue to represent the most scalable mechanism for bypassing MFA at enterprise scale, with Tycoon 2FA responsible for 62% of Microsoft-blocked phishing attempts before Europol disruption in March 2026. The fundamental limitation of all AiTM defenses based on endpoint detection is that session cookie theft occurs server-side through the reverse proxy interception of authenticated sessions—making the attack invisible to endpoint security controls that monitor for credential theft. WhatsApp's publication of three IOCs from NSO Group's renewed spear-phishing campaign (ikhwancast[.]com, ghazacast[.]com, fr24cast[.]com), targeting high-risk individuals in Jordan and Lebanon despite a permanent injunction, illustrates the persistence of identity-targeting campaigns against specific high-value individuals where the threat actor calculates that targeted surveillance value exceeds legal and reputational risk.

AI-driven phishing is creating structural strain on SOC Tier 1 operations, with Microsoft's 2025 Digital Defense Report documenting AI-automated phishing achieving a 54% click-through rate versus 12% for standard attempts—a differential that simultaneously increases the volume of security alerts requiring investigation and degrades the signal quality of existing behavioral detection rules trained on lower-sophistication phishing patterns. The FIFA World Cup 2026 threat campaign, with Recorded Future tracking 33 themed purchase scam domains connected to approximately 2,500 Meta advertisements, illustrates how major public events are systematically exploited for credential harvesting at scale using AI-generated content. Defensive priorities should emphasize phishing-resistant FIDO2/hardware token MFA deployment, domain binding and device-based session validation controls that resist AiTM interception, and behavioral analytics capable of detecting post-authentication anomalies that indicate session token compromise rather than relying on pre-authentication controls that motivated attackers have demonstrably circumvented.

The University of Oxford's disclosure of a data breach affecting its CareerConnect platform via third-party vendor Group GTI marks the institution's second reported breach of 2026 and exemplifies the systemic risk posed by third-party service providers with privileged access to sensitive university user data. SoFi's Hong Kong subsidiary similarly confirmed customer data exposure through a third-party vendor database compromise, reinforcing the pattern of supply chain and vendor relationship exploitation as a primary initial access vector for data theft. The Lansing Community College breach affecting 174,000 individuals continues a trend of educational sector targeting, while the BGF Networks CU parcel delivery service breach in South Korea—now subject to a police inquiry—demonstrates the geographic breadth of data theft operations targeting consumer services platforms with aggregated personal information.

At the geopolitical and criminal nexus, the PSNI data breach conviction of Christopher Paul O'Kane for terrorist offences represents a significant legal precedent for the criminal weaponization of accidentally disclosed government personnel data. O'Kane's systematic manipulation of leaked officer information into targeted spreadsheets demonstrates that even inadvertent data disclosures carry immediate national security implications when the exposed population includes law enforcement personnel. The Meta Mercor breach—where TeamPCP's exploitation of LiteLLM supply chain vulnerabilities enabled Lapsus$ to subsequently publish 4TB of stolen data from an AI data startup serving Meta, OpenAI, and Anthropic—illustrates an emerging category of high-impact breaches where a single compromise of a shared third-party supplier simultaneously exposes multiple competing organizations in the AI development ecosystem, with systemic rather than isolated organizational impact.

On the detection and intelligence front, several meaningful advances in defensive tooling and methodology have been released. The open-source 'CTI as a Code' framework introduces version-controlled, reproducible threat intelligence operations with four operational modes, evidence confidence labeling, and automatic Sigma rule generation with MITRE ATT&CK mapping integrated across tools including OpenCTI, TheHive, Elastic, and Velociraptor. Microsoft Defender for Endpoint has added RPC activity monitoring, addressing one of the most persistently abused lateral movement vectors in Windows environments. Cloudflare's new integration enabling real-time WAF rule generation from threat intelligence data—automatically blocking traffic attributed to known actors like Tycoon 2FA and RaccoonO365—represents a meaningful reduction in the manual effort required to operationalize threat feeds into preventive controls.

The Qilin ransomware collective continues to demonstrate outsized operational tempo, with ZeroFox assessing the group as the most active ransomware operation globally in Q2 2026 following 15 new claimed victims across nine countries between June 2–5 alone. The group's exploitation of the Check Point VPN zero-day for approximately one month before a patch existed exemplifies the sustained advantage threat actors maintain when targeting network access infrastructure. Defenders should note the broader trend identified by Kaspersky researchers linking hacktivist groups including 4BID and Goffee to financially motivated expansion beyond political targets, exploiting ProxyShell (CVE-2021-34473) for initial access and deploying a diverse toolset including BlackReaperRAT and ClearWater ransomware—a convergence of hacktivist and criminal tradecraft that complicates attribution and response prioritization.

Vendor security advisories this period reveal significant vulnerability density across enterprise cloud infrastructure components. Broadcom disclosed three VMware Cloud Foundation stored XSS vulnerabilities (CVE-2026-41722 through CVE-2026-41724, CVSS 8.0) affecting Cloud Foundation, vSphere Foundation, Aria Operations, and Telco Cloud Platform, enabling authenticated attackers with object-creation privileges to inject scripts that execute in administrator contexts. A comprehensive IBM security bulletin (AV26-553) addresses critical vulnerabilities across 40+ products including Cloud Pak for Data, QRadar, WebSphere Application Server, and Sterling Connect:Direct, reflecting the breadth of enterprise cloud platform exposure from a single vendor's disclosure cycle. Dell advisories covering Private Cloud-VMware, PowerSwitch, and VxRail Appliance add further patching burden for organizations operating hybrid on-premises and cloud infrastructure.

Container security remains structurally problematic at scale, with Kaspersky research revealing that only 10% of the most downloaded Docker images are secure, 64% contain critical vulnerabilities including remote code execution and privilege escalation flaws, and many are maintained by volunteers who cease updates—leaving months or years of unpatched vulnerabilities in widely deployed production container images. The Redis DarkReplica vulnerability (CVE-2026-23631), exploiting use-after-free conditions in the Lua VM through the SLAVEOF replication mechanism to achieve authenticated RCE with full server compromise, is particularly significant given Redis's ubiquity as a cloud-native caching layer. Aviatrix's integration with the OISF Suricata project to advance open-source threat detection within cloud environments, and Tenable's cloud detection and response capabilities emphasizing near-real-time threat signal unification and identity context, represent the defensive tooling maturation needed to address these compound cloud security challenges.

The NSO Group's continued defiance of legal orders represents the most consequential commercial spyware development of the period. WhatsApp detected and disrupted new spear-phishing campaigns attributed to NSO using malicious links to redirect users to external sites for Pegasus infection, and has filed a contempt motion publishing three IOCs (ikhwancast[.]com, ghazacast[.]com, fr24cast[.]com) used in the operation. The campaign targeted high-risk individuals including journalists, activists, and diplomats in Jordan and Lebanon, targeting fewer than 10 users with no confirmed successful compromises—though the surveillance-for-hire model's persistence despite a $168 million court judgment and U.S. Entity List placement indicates that commercial deterrence through legal and regulatory mechanisms alone is insufficient. Google's Fake Call Detection rollout for Android 12+ devices, using encrypted RCS cryptographic handshakes to verify caller device authenticity and detect AI voice-cloned impersonation, addresses a different dimension of the mobile threat surface: the $2.95 billion in FTC-reported U.S. impersonation scam losses in 2024 that deepfake voice technology is amplifying.

Apple's emergency iOS patches for active web-based exploitation campaigns targeting devices running iOS 15, iOS 16, and earlier versions—where attackers exploit browser rendering vulnerabilities to bypass the application sandbox and access system directories—demonstrate that legacy iOS support creates a persistent attack surface even as newer devices receive iOS 26 hardening. The UC San Diego team's disclosure and subsequent patching of a critical SMS spoofing vulnerability affecting Android, iOS, and all major U.S. carriers including Verizon reflects the systemic risk created by the inconsistent email-to-SMS gateway translation protocols that were never engineered for secure interoperability. The vulnerability's exploitation through special characters in email addresses to forge sender identities and inject messages into existing conversations represents a social engineering enabler of broad applicability, and its resolution required coordinated action across carriers that historically operate independent SMS infrastructure.

Cross-chain bridge security continues to be the highest-impact attack surface in the DeFi ecosystem, with the Syscoin bridge exploit of June 7 demonstrating how implementation errors in SPV proof validation logic—specifically a parsing error that incorrectly interpreted malformed proofs as valid—can enable unauthorized minting of billions of tokens (approximately 5 billion SYS, valued at ~$10 million) without corresponding burn transactions. The April KelpDAO hack's exploitation of a LayerZero bridge vulnerability for approximately $292 million, which triggered an $8.45 billion bank run on Aave over 48 hours and required a $300 million emergency bailout, illustrates the systemic interconnectedness risk where a single bridge compromise can cascade through multiple dependent protocols. The Humanity Protocol private key compromise—resulting in $30+ million in token theft and an 88% price collapse within 12 hours—reflects a different but equally consequential attack vector: the persistent risk of private key compromise in cryptocurrency projects where single points of cryptographic failure exist without multi-signature or threshold signature controls.

North Korean cryptocurrency theft operations remain an active structural threat to the ecosystem, with UNK_DeadDrop sending over 250 targeted emails to developers at approximately 100 organizations using fake developer recruitment and code review lures to distribute cross-platform credential theft and cryptocurrency wallet compromise malware. The campaign's tactical shift from LinkedIn to direct email and from fake interviews to unsolicited job offers reflects operational adaptation following increased security community awareness of the Contagious Interview tradecraft. A broader pattern of business logic exploitation—exemplified by the Osmosis/Ethereum bridge attack where an attacker minted worthless tokens and embedded legitimate ETH custody addresses within fabricated denomination strings before calling a permissionless register() function to release real funds—indicates that the DeFi threat surface increasingly requires adversarial modeling of cross-chain economic logic rather than exclusively smart contract code review.

The Microsoft GitHub repository compromise, affecting over 73 repositories across four organizations including Azure and Azure-Samples, is the highest-profile supply chain incident of the period and raises systemic questions about contributor account security and the trust model underlying open-source AI tool ecosystems. Microsoft's characterization of the incident as a 're-compromise' of the Durable Task ecosystem—with attackers apparently retaining credentials from a May breach—illustrates how initial supply chain compromises create persistent access vectors that enable cascading follow-on attacks. GitHub's disabling of affected repositories and Microsoft's emergency notification of a small number of affected customers represent reactive containment measures; however, the Miasma worm's self-replicating credential harvesting architecture means that any developers who opened affected tools in AI coding applications may have had credentials silently exfiltrated to GitHub C2 repositories before detection.

Defensive responses across the ecosystem are beginning to mature, though structural gaps remain. Microsoft's two-hour VS Code extension update delay provides a detection window for obvious malicious extensions but offers limited protection against validly-signed packages from compromised publisher accounts—the primary vector in the current campaign. The Lazarus Group's continued npm brandjacking campaign, employing suffix-addition, version mimicry, and embedding tactics beyond simple typosquatting to impersonate Buffer, Chai, React, and webpack, illustrates that nation-state actors are operating in the same package ecosystem as criminal supply chain attackers. Supabase's documentation of defensive controls including minimum 7-day release age quarantine, provenance signature verification, lifecycle script constraints with default-deny policies, and SHA512-pinned package manager versions represents the emerging best practice standard—though adoption across the broader developer community remains far from universal. The OpenSSF's warning about 66% CRA unawareness among manufacturers and developers operating in the EU reinforces that regulatory frameworks intended to mandate supply chain security controls face significant awareness and implementation gaps before they can produce measurable risk reduction.

The politicization of deepfake technology is generating both immediate fraud harm and longer-term democratic integrity concerns. A Russian-speaking fraud group weaponized AI-generated videos of UK political figures Nigel Farage and Bank of England governor Andrew Bailey in a coordinated investment scam on X (Twitter), exploiting the platform's reduced content moderation capacity to reach victims before takedown. A deepfake attack campaign targeting U.S. swing voters documented 3,124 verified incidents with 412% month-over-month growth, reaching 14.2 million voter impressions before takedown—with 65% using voice-cloned content specifically targeting the segment of the electorate most susceptible to persuasion. South Korean police launched a criminal investigation into deepfake video production used to defame a governor election candidate in violation of the Public Official Election Act, reflecting the leading edge of legal enforcement frameworks catching up with electoral deepfake deployment.

Defensive countermeasures are beginning to achieve technical parity in specific attack vectors. Google's Fake Call Detection for Android 12+ uses end-to-end encrypted RCS cryptographic handshakes to verify caller device authenticity—a device-to-device attestation approach architecturally distinct from and more robust than STIR/SHAKEN carrier-level authentication, which can be bypassed by attackers with carrier access. OCR Studio's neural network for detecting document collages and AI-generated ID forgeries in KYC workflows addresses the identity verification attack surface directly, analyzing local visual anomalies and algorithmic artifacts from models including ChatGPT, Midjourney, and Grok to expose high-quality forgeries. However, the fundamental challenge remains: deepfake detection capability is inherently reactive to generation capability, and as generation models continue to improve in acoustic naturalness, visual fidelity, and temporal coherence, detection systems face a persistent capability gap that organizational verification protocols—including secret family code words and mandatory callback verification through independent channels—must compensate for through procedural rather than purely technical controls.

A critical red team tool development merits immediate defensive attention: the release of EDRChoker, which exploits Windows Policy-Based Quality of Service to throttle EDR process bandwidth to approximately 8 bits per second via the pacer.sys NDIS Lightweight Filter Driver. Unlike previous EDR evasion tools that leverage the Windows Filtering Platform API and generate detectable packet-drop alerts, EDRChoker causes connection timeouts rather than blocking events at 8 bps—a bandwidth level that makes standard TLS handshakes impossible—specifically bypassing detection rules monitoring for WFP-based evasion. This technique effectively isolates EDR agents from cloud command infrastructure without generating firewall events, representing a significant addition to the post-exploitation toolkit that defenders must account for in detection engineering. The publication of this technique as open-source means widespread adversarial adoption should be anticipated in the near term.

The post-quantum cryptography preparation timeline is receiving renewed urgency from Germany's ANSSI and other national cybersecurity agencies, with experts emphasizing that the transition from RSA and ECC to post-quantum algorithms is a multi-year infrastructure undertaking requiring immediate planning even without confirmed Q-Day timeline. WISeKey, Hashgraph, and Hedera's launch of the QAIT Q-Day Security Assessment Platform on the SEALCOIN Quantum Marketplace represents the commercialization of quantum readiness assessment, enabling organizations to identify cryptographic vulnerabilities and generate migration roadmaps using AI-assisted analysis. The SEI and Accenture's empirically based AI Adoption Maturity Model, designed to help organizations scale AI with predictable security outcomes across a rapidly changing technology landscape, addresses a parallel governance challenge: as AI tools proliferate faster than security teams can assess them, maturity models provide the structured evaluation frameworks necessary for risk-informed deployment decisions.

The EU Cyber Resilience Act continues to generate significant compliance preparedness gaps, with OpenSSF reporting that 66% of global manufacturers and developers remain unaware of the regulation despite a December 2027 compliance deadline. The finding that 41% of affected organizations have not determined whether the CRA applies to them, and only 32% produce SBOMs for all products, indicates that the open-source ecosystem—where 62% of SMEs rely on open-source for over 75% of their products—faces a structural readiness deficit with compounding technical debt averaging $258,000 per release cycle. ENISA's release of Technical Competence Requirements for CRA Notified Bodies in June 2026 represents the regulatory infrastructure maturation needed to operationalize the act, but the awareness gap suggests enforcement will encounter significant market friction.

Human risk and behavioral governance are receiving renewed regulatory and audit attention following the 2026 Verizon DBIR finding that humans were involved in 62% of breaches. The emerging tension between 'Human Risk Management' and 'Secure Behavior Management' frameworks reflects a broader shift in regulatory expectations: auditors are increasingly requiring demonstrable behavioral outcomes—reporting velocity, incident prevention rates, cost avoidance metrics—rather than traditional participation metrics such as training completion rates and phishing simulation click rates. This evolution in measurement standards creates near-term compliance pressure for organizations whose security awareness programs are built around activity reporting rather than behavioral change validation, particularly as AI-crafted phishing threats render historically acceptable simulation benchmarks inadequate proxies for actual employee resilience.

Critical infrastructure targeting is expanding geographically and sectorally, with Bernard Chadenga's warning that African critical infrastructure is increasingly targeted by nation states, activist groups, and cybercriminals reflecting a global broadening of ICS threat actor interest beyond traditional Western targets. The cited Florida water treatment facility incident—where attackers altered chemical treatment levels to dangerous maximums with disaster averted only by manual employee intervention—serves as the paradigmatic illustration of why OT security failures carry physical consequence dimensions absent from enterprise IT incidents. The White House Executive Order 14409's explicit prioritization of critical infrastructure protection and CISA's CI Fortify doctrine requiring critical infrastructure to survive cyber isolation represent the policy response to this threat trajectory, though the gap between policy intent and operational implementation across diverse infrastructure sectors remains significant.

Industry responses include both commercial tool expansion and novel research contributions. Honeywell's OT Cybersecurity Suite expansion adds AI-powered Cyber Proactive Defense with real-time monitoring and analytics, a Secure Media Exchange portable scanner for air-gapped environments, and expanded 24/7 managed SOC services—addressing the finding that only 32% of industrial organizations actively monitor OT systems and 20% maintain dedicated OT security teams. University of Malaga researchers have demonstrated a multi-agent security system using consensus mechanisms and blockchain technology for EV charging station attack detection, addressing the expanding attack surface of electrification infrastructure. The railway cybersecurity market's projected growth from $6.70 billion to $11.55 billion by 2031 reflects broader capital allocation toward OT security as critical infrastructure operators internalize lessons from high-profile incidents across energy, water, and transportation sectors.