State of Security | February 2026
MustardTree Partners Monthly Cybersecurity Report
February 2026 has been defined by convergence. The threats we’ve been tracking individually — AI-powered offensive operations, nation-state supply chain infiltration, ransomware fragmentation, and regulatory acceleration — are no longer operating in parallel. They’re colliding. The Milan-Cortina Winter Olympics provided a live demonstration of what happens when geopolitical tension meets digital infrastructure at scale. Meanwhile, an unprecedented wave of zero-day exploitation forced emergency responses across every major platform simultaneously, and the cybersecurity industry’s own structural transformation accelerated with the largest acquisition in its history. This month’s report examines the forces reshaping the threat landscape and what they demand of defenders.
The Olympics Under Siege: When Cyber Meets Geopolitics
The Milan-Cortina 2026 Winter Olympics became the most heavily targeted Games in history — and the most extensively defended. Italy’s National Cybersecurity Agency stood up a 24/7 command centre in Rome, deploying 6,000 security officers across venues spanning 22,000 square kilometres of northern Italy. They needed every one of them.
Pro-Russian hacktivist group NoName057(16) launched sustained DDoS campaigns against Olympic infrastructure, Italian government websites, and the diplomatic network — including the Italian embassy in Washington and consulates across four continents. The group framed the attacks as retaliation for Italy’s support of Ukraine, but the operational pattern reveals something more calculated. DDoS was the visible layer; the concern among security teams was that volumetric attacks served as cover for deeper intrusion attempts against operational technology: power grids in the Dolomites, snow-making systems, scoring networks, and ticketing platforms.
The threat landscape extended well beyond Russia. Palo Alto Networks’ Unit 42 identified APT28, China’s Mustang Panda, and North Korea’s Kimsuky as credible threats. The BD Anonymous group announced an #OpItaly campaign, while Z-Pentest Alliance and Server Killers claimed attacks against Italian industrial control systems. Russia’s exclusion from the Games — driven not by doping disputes but by the geopolitical fallout of the Ukraine invasion — removed a critical restraining influence. When marquee winter sports like ice hockey and figure skating are absent from the Russian national conversation, the calculus around offensive operations shifts.
The Milan-Cortina experience confirms what security strategists have long warned: major international events are no longer just physical security challenges. They are cyber battlegrounds where nation-state proxies, hacktivists, and opportunistic criminals converge simultaneously.
Zero-Day Avalanche: February’s Emergency Patching Crisis
February 2026 will be remembered as an emergency-level patching month. Microsoft’s Patch Tuesday addressed 58 vulnerabilities, including a staggering six actively exploited zero-days — an event that prompted CISA to add all six to its Known Exploited Vulnerabilities catalogue with a March 3rd remediation deadline for federal agencies.
The severity profile is alarming. CVE-2026-21510, a Windows Shell bypass rated CVSS 8.8, allows a single click on a malicious link to silently execute attacker-controlled content without any warning dialog. CVE-2026-21513 achieves similar results through the MSHTML Framework. CVE-2026-21533, discovered by CrowdStrike, revealed that threat actors had been exploiting a privilege escalation flaw to target organisations in the United States and Canada since at least December 2025 — meaning defenders were unknowingly exposed for months before the patch arrived.
But Microsoft wasn’t alone. Google patched CVE-2026-2441, the first actively exploited Chrome zero-day of the year — a use-after-free vulnerability in CSS that enables arbitrary code execution through a malicious website. Apple shipped emergency updates across its entire ecosystem for CVE-2026-20700, a flaw weaponised in what the company described as an “extremely sophisticated attack” targeting specific individuals. BeyondTrust disclosed CVE-2026-1731 with a CVSS score of 9.9, actively exploited in the wild. Dell’s RecoverPoint for Virtual Machines was found to contain hard-coded credentials enabling unauthenticated remote root access.
The simultaneous exploitation across Microsoft, Google, Apple, Dell, and BeyondTrust products represents a fundamental challenge: organisations cannot triage when everything is critical simultaneously. The traditional “patch by severity” approach collapses when every major vendor releases emergency fixes in the same window.
The Shadow Campaign and the Supply Chain Siege
Nation-state operations reached a new threshold of ambition in February. Palo Alto Networks disclosed what it calls the “Shadow Campaign” — a state-sponsored espionage operation that compromised at least 70 organisations across 37 countries. Tracked as TGR-STA-1030 and assessed with high confidence to be a Chinese nexus group, the campaign targeted government agencies and critical infrastructure with systematic precision.
The Notepad++ supply chain compromise may prove even more consequential. Between June and December 2025, the Lotus Blossom group — a known state-sponsored threat actor — infiltrated the official hosting infrastructure for Notepad++, one of the world’s most widely used text editors. They intercepted and redirected traffic destined for the update server, selectively targeting users in Southeast Asian government, telecommunications, and critical infrastructure sectors. The implications are stark: if a ubiquitous open-source tool’s update mechanism can be silently weaponised for six months, the entire software supply chain model requires re-examination.
Meanwhile, the FBI confirmed at CyberTalks that Salt Typhoon remains an active, ongoing threat. The Chinese espionage group’s infiltration of US telecommunications infrastructure — now confirmed at nine major carriers — continues to provide counterintelligence capabilities that effectively wiretap the wiretappers. Volt Typhoon’s pre-positioning in critical infrastructure for potential kinetic conflict scenarios persists, with renewed attempts to re-establish access to networks from which it was previously evicted. Ivanti vulnerabilities were exploited in targeted attacks against the European Commission and Dutch and Finnish government agencies.
The common thread across these operations is patience. These are not smash-and-grab campaigns. They are systematic, long-duration infiltrations designed to persist undetected while providing strategic advantage.AI Crosses the Rubicon: From Tool to Autonomous Attacker
The transition from AI-assisted to AI-driven offensive operations accelerated dramatically. A Russian-speaking threat actor leveraged commercial generative AI services to compromise over 600 FortiGate devices across 55 countries between January and February. The significance isn’t the scale alone — it’s that the attacker was assessed as unsophisticated. AI transformed weak credentials and exposed management ports into a global campaign that would previously have required nation-state resources.
The agentic AI threat moved from theoretical to operational. Anthropic reported an observed intrusion where AI executed 80–90% of the activity autonomously, with human operators intervening only at a handful of decision points. This represents a fundamental inflection: the adversary’s constraint is no longer technical skill but access to AI tooling.
On the defensive side, the enterprise rush to deploy AI agents created its own attack surface. A Dark Reading poll found 48% of cybersecurity professionals now identify agentic AI as the top attack vector heading into 2026. The WEF’s Global Cybersecurity Outlook reports 94% of respondents believe AI will be the single biggest driver of change in cybersecurity this year. The risks are specific and documented: prompt injection attacks achieving 92% success rates across open-weight models, agent-to-agent impersonation enabling unauthorised capability escalation, memory poisoning of long-term agent storage, and supply chain attacks through the rapidly adopted Model Context Protocol ecosystem.
NIST responded by launching its AI Agent Standards Initiative, acknowledging that more than 80% of Fortune 500 companies now deploy active AI agents. The gap between deployment speed and security maturity is widening — and adversaries are already exploiting it.
The Ransomware Metamorphosis
The ransomware ecosystem continued its structural transformation. Attack volume surged — publicly reported incidents rose 47% year-over-year to approximately 7,200 in 2025, with February showing no signs of deceleration. The FCC cited a fourfold increase in ransomware attacks since 2021 in urging telecommunications firms to strengthen defences.
But the economics are shifting beneath the surface. Ransom payments declined in both total volume and average size, as more organisations refuse to pay, invest in backup resilience, and engage law enforcement. Sixty-four percent of victims now refuse payment on principle. The response from threat actors has been predictable: the pivot to pure data extortion accelerated. Encryption is increasingly bypassed entirely in favour of exfiltrating sensitive data and threatening public release. This reduces technical overhead while maintaining leverage through regulatory exposure and reputational damage.
The ecosystem’s fragmentation intensified. Where 2025 still featured recognisable ransomware brands, February 2026 saw rapid rebranding, affiliate migration, and the emergence of new groups including CipherForce and NightSpire alongside established operators like Qilin and ShinyHunters. Attribution has become significantly harder. Recorded Future assessed that 2026 will be the first year the number of new ransomware operators outside Russia exceeds those emerging within it — reflecting global expansion rather than Russian decline.
The insider threat dimension grew more concerning. Ransomware operators are increasingly recruiting corporate insiders — specifically targeting native English speakers — to provide initial access. With credential-based intrusions already dominating the initial access landscape, the combination of purchased credentials, vulnerability exploitation, and recruited insiders creates a multi-vector entry problem that perimeter defences alone cannot address.
Google-Wiz: The $32 Billion Bet Reshapes the Industry
On February 10th, the European Commission granted unconditional approval for Google’s $32 billion acquisition of Wiz — the largest cybersecurity acquisition in history. With prior DOJ clearance, the deal is now positioned to close, fundamentally reshaping the competitive landscape.
The strategic logic is clear: the integration of Wiz’s cloud security scanning capabilities with Google’s Gemini AI models and Mandiant’s threat intelligence creates an end-to-end autonomous security platform. Google’s vision of AI-native security operations — where scanning, detection, and response happen at machine speed — moves from concept to operational capability.
The competitive implications are significant. While Google committed to maintaining Wiz’s cross-platform support for AWS, Azure, and Oracle Cloud, the potential for preferential integration with Google Cloud is the elephant in the room. For organisations building multi-cloud security architectures, the consolidation raises questions about vendor neutrality and long-term platform strategy. For the broader market, it signals that cybersecurity’s next phase will be defined by whoever best integrates AI, cloud infrastructure, and threat intelligence into a unified platform.
Regulatory Momentum Builds Across Jurisdictions
The regulatory landscape tightened on multiple fronts. The EU’s NIS2 Directive moved decisively from national transposition to active enforcement, significantly expanding the number of organisations in scope. The Cyber Resilience Act’s mandatory security requirements for connected products approach their 2026 compliance deadline. The European Cybersecurity Act’s proposed revision reflects a 150% increase in cyber-attacks since its original adoption, with expanded scope and strengthened requirements.
In the United States, the regulatory picture remained complex. The CISA incident reporting rule was delayed until May 2026 amid industry criticism that the proposed requirements are overly broad. CISA itself enters 2026 without a Senate-confirmed director, creating leadership uncertainty at a critical moment. At the state level, the patchwork expanded to 20 states enforcing consumer privacy statutes, with California continuing to refine requirements around automated decision-making and cybersecurity audits. New York’s Department of Financial Services began its first full examination cycle under amended cybersecurity regulations, with enforcement focus on governance, risk assessment, and multi-factor authentication.
The US House passed the PILLAR Act to renew cybersecurity grants for state and local governments — an acknowledgement that municipal and state-level organisations face nation-state calibre threats with significantly fewer resources. Smaller financial institutions face a June 3rd compliance deadline under amended Regulation S-P requirements, including new mandates for incident response programmes, customer notification, and service provider due diligence.
The overarching trend is regulatory convergence toward accountability: organisations must demonstrate not just compliance but active, measurable security posture improvements.
Strategic Imperatives
February’s events demand specific defensive responses:
Adopt crisis-speed patching protocols. The simultaneous exploitation across multiple vendors requires pre-approved emergency patching workflows that bypass standard change management timescales. Waiting for the next maintenance window is no longer viable when six zero-days are actively exploited.
Harden the software supply chain. The Notepad++ and AI model supply chain compromises demonstrate that trusted software distribution channels are now primary attack vectors. Implement integrity verification for all software updates, restrict auto-update mechanisms in sensitive environments, and maintain software bills of materials.
Secure AI agent deployments. The rush to deploy agentic AI is outpacing security controls. Every AI agent requires identity governance, privilege boundaries, output validation, and monitoring equivalent to a human employee with the same access. Memory persistence, tool access, and inter-agent communication are all attack surfaces that demand immediate attention.
Prepare for data-only extortion. Traditional ransomware defences focused on backup and recovery are insufficient when attackers skip encryption entirely. Data loss prevention, network segmentation, and exfiltration detection must be prioritised alongside — not behind — backup resilience.
Consolidate event-driven threat intelligence. Major international events are now predictable triggers for coordinated cyber campaigns. Organisations with any connection to host nations, participating organisations, or supporting infrastructure must integrate event calendars into their threat modelling.
Looking Ahead
March will bring the conclusion of the Milan-Cortina Games and a likely surge in post-event disclosure of attacks that were contained but not yet public. The CISA incident reporting rule’s progress toward its May deadline will shape US reporting obligations for years to come. The Google-Wiz integration will begin revealing whether the promise of AI-native security operations can be delivered at scale. And the ransomware ecosystem’s fragmentation will continue testing attribution capabilities and law enforcement coordination.
The thread connecting February’s events is acceleration. Threats are moving faster, exploits are being weaponised sooner, AI is lowering barriers more rapidly, and the regulatory response — while heading in the right direction — is struggling to keep pace. The organisations that thrive will be those that match this acceleration with their own: in detection, in patching, in adaptation, and in strategic foresight.
This report is produced by MustardTree Partners as part of our ongoing commitment to providing actionable cybersecurity intelligence for senior leadership and technology decision-makers
.