Why the Threat Changed
Mobile banking trojans like Mamont, Coper, and Rewardsteal are proliferating, with Mamont accounting for nearly half of all banking trojan attacks in 2025. These are active campaigns rewriting the economics of mobile intrusion.
The phone in your pocket has become the most reliable way into your employer's network. Not because the device is weak, but because the people holding it are easier to trick than any firewall. Signature-based scanners — the workhorse of mobile security for a decade — are losing relevance as attackers move from mass-distribution trojans to targeted, AI-assisted campaigns that leave no file to scan. A gap has opened between what legacy tools catch and what actually reaches the device. Enterprises are noticing.
Kaspersky's telemetry recorded 14 million blocked attacks involving malware, adware, or unwanted mobile software across 2025, driven by roughly 1.17 million attack attempts per month. Adware still dominates by volume at three in five detections, but the sharper signal is in banking trojans: 255,000 new unique installation packages appeared last year, a several-fold increase over prior years, and total banker attacks grew by half. The same report logged 815,000 new malicious packages overall, a decline in raw count, but a rise in sophistication. Threat actors are diversifying delivery channels and accelerating variant production to evade detection, Kaspersky reported in its March 2026 mobile threat report.
The economics explain the pivot. Android holds roughly seven in ten phones globally, and its openness makes it a scalable target. In the second quarter of 2024 alone, researchers counted 367,000 new Android malware packages. Iran ranks as the most-impacted country by mobile malware, while the United States sees nine times more attacks annually than the United Kingdom. Seven in ten organizations have users being served malware-laden ads on their browsers, a vector that bypasses app-store vetting entirely.
Kimwolf goes after Android TV boxes. LunaSpy masquerades as antivirus software. Coper banking trojans and Hqwar droppers concentrate in Türkiye; Rewardsteal trojans proliferate in India under the guise of monetary giveaways; a Trojan-Proxy campaign in Germany abused a compromised retail discount app to turn victim devices into residential proxies; Pylcasa trojans in Brazil redirect users to phishing pages and illicit casino sites. Each campaign uses different lures, but all share a pattern: they exploit trust, not just code flaws.
The window between a vulnerability's discovery and its active exploitation has collapsed to hours. Attackers probe the average application every four minutes. The most frequent confirmed exploit is untrusted deserialization, a complex flaw allowing malicious code injection through serialized data. Applications under observation typically carry 106 identified vulnerabilities in custom code, 22 of them high or critical severity. Remediating a critical vulnerability takes 92 days on average; the monthly remediation rate is just 3.4 vulnerabilities per application. The math does not work for manual patching.
Generative AI accelerates the offense. Adversarial samples crafted to deceive deep-learning classifiers bypass image-based malware detection. Attackers now produce variants at a pace that outstrips signature generation. The same AI tooling that helps developers write code helps threat actors obfuscate payloads, generate phishing lures, and automate reconnaissance.
The impact is measurable. Four in ten malware attacks result in confidential data leakage. Over seven in ten malware attacks target a specific victim. The average ransom payment reached $1.5 million in 2023, nearly double the 2022 figure. Nearly three in four businesses worldwide were affected by ransomware as of 2023. IBM puts the average time to identify a ransomware attack at 49 days. Phishing costs $4.9 million per incident. Eight in ten data breaches involve cloud-based data. Half of organizations experienced a ransomware event that significantly disrupted operations; five in six paid the ransom, and over half paid more than $100,000.
Traditional mobile device management and basic endpoint protection were built for a different threat model — one where malware arrived as a file, where the perimeter was definable, and where patching cycles measured in weeks were acceptable. None of those assumptions hold. The attack surface has moved to the identity layer, the browser, and the app supply chain. Defending it requires detection that does not depend on prior knowledge of a sample.
That requirement is pushing enterprises toward mobile threat defense platforms that embed behavioral AI, systems that model normal device, network, and application behavior and flag deviations in real time. The next section examines how Lookout, Zimperium, and Microsoft are shipping those capabilities, and where the procurement signal is strongest.
Lookout's AI Governance Play
Lookout rolled out AI Visibility & Governance in 2024 as a dedicated module inside its Mobile Endpoint Security platform. The release targets a gap the company has been documenting for months: enterprise AI activity increasingly lives on smartphones and tablets, outside the firewalls, secure web gateways, and endpoint agents that traditional governance tools monitor. Employees now interact with generative AI apps, embedded AI features, AI-powered assistants, and early agentic workflows from mobile devices. That traffic bypasses corporate networks, traditional endpoint controls, secure web gateways, and existing AI governance monitoring tools, creating what Lookout calls a "dangerous illusion of oversight."
The module delivers six concrete capabilities: AI application discovery and inventory, shadow AI visibility, monitoring of AI-enabled mobile applications, governance and policy enforcement for AI usage, visibility into agentic AI behaviors and data flows, and audit-ready evidence supporting AI governance programs.
The question is no longer: 'Can this platform detect mobile threats?' The more important question is: 'Can this platform provide visibility, governance, and control over enterprise AI usage occurring across mobile environments?'
The engine behind those features is the Lookout Security Cloud, an AI-driven dataset the company says spans more than 220 million devices, 325 million apps, and billions of web items. Since 2019 it has cataloged 465 million phishing and malicious sites. That telemetry feeds models that classify AI apps, score their risk, and correlate mobile AI activity with broader threat signals — phishing campaigns, malware families, command-and-control infrastructure.
Lookout shipped AI Visibility & Governance alongside two proactive protections aimed at the social-engineering surge the dataset is measuring. Executive Impersonation Protection compares inbound SMS senders against a customer's executive directory and alerts when an unknown number claims to be the CEO or CFO. Advanced Smishing Protection inspects links in messages across any app and warns the user before the tap; if the user taps anyway, the URL is blocked at the network layer. Both features run on iOS and Android. Admin Visibility and Reporting aggregates those alerts in the security console so teams see coordinated campaigns as they unfold, not weeks later in a forensic review.
IDC's Mike Jude, research director for endpoint security, framed the shift: "Mobile devices play a pivotal role in the enterprise because they enable remote access to cloud apps and data. When left vulnerable to phishing and social engineering attacks, they expose a critical weakness within any organization." Lookout's own telemetry backs the urgency. The Q2 2024 Threat Landscape Report recorded a 70 percent year-over-year increase in mobile phishing and malicious web content and a 40 percent rise in enterprise mobile phishing attempts and malicious web attacks.
Firas Azmeh, president of Mobile Endpoint Security at Lookout, said the release came from direct customer demand: "In recent conversations with customers and industry analysts, we've been told that smishing and executive impersonation are two of the most frustrating security challenges to deal with. Knowing your organization is under attack from a targeted threat is critical for today's security teams, so by combining all three of these solutions into one release, we're helping our customers know as soon as their employees are being targeted en masse so they can prevent phishing incidents, mitigate damage and protect organizational data."
The timing aligns with procurement cycles in regulated sectors. Government and financial-services buyers now evaluate mobile security platforms against AI governance checklists — discovery, shadow AI visibility, policy enforcement, agentic behavior monitoring, audit evidence — alongside traditional MTD criteria like device compromise detection and malware blocking. Platforms built on legacy threat paradigms risk immediate capability gaps that require supplemental tooling or re-procurement. Lookout's bet is that AI Visibility & Governance becomes a primary buying criterion, not an add-on, and that the 16-year MTD track record plus the Security Cloud's scale give it a defensible lead in that evaluation.
Zimperium and Microsoft: Different Paths, Same Gap
Zimperium has spent more than a decade building AI into the foundation of its mobile security stack rather than layering it on top. The company's founding CTO, Itzhak "Zuk" Avraham, brought three years of Israeli Defense Forces experience to the problem set when he incorporated Zimperium in 2010, and the first commercial product — zIPS for Android — shipped in 2014 with on-device machine learning that monitored user behavior to detect compromise. That early architectural bet matters now: the current MTD platform runs its dynamic detection engine entirely on the device, so it can identify zero-day malware, phishing from any vector (SMS, WhatsApp, Messenger), network attacks, and app vulnerabilities without cloud round-trips or signature updates. Over-the-air protection updates keep defenses current without requiring app resubmissions, a practical advantage for enterprises managing fleets across Android, iOS, and ChromeOS.
The platform's dual AI architecture distinguishes it from competitors who treat AI as a console-side analytics layer. One AI agent lives inside the mobile app, performing real-time detection and response at the endpoint; a second, the Mobile SOC Agent, runs on the management console, giving security, fraud, and SOC teams a shared investigation workspace the moment a threat surfaces. Zimperium formalized this approach in May 2026 with the Mobile App Response Agent, explicitly designed to arm security and fraud teams against the rising tide of AI-generated mobile attacks. The company's MAPS platform extends the same philosophy to application protection, unifying shielding, runtime protection, security testing, and cryptographic key management into a single ecosystem that serves financial services, healthcare, and government customers. QKS Group recognized this integration in June 2026, positioning Zimperium as a Leader in its SPARK Matrix for In-App Protection Platforms.
Independent validation has followed. In July 2024, an independent research firm awarded Zimperium MTD the highest possible score across 17 criteria, including Application Integrity, Real-Time Threat Detection, Phishing Defense, Malicious App Prevention, and Vulnerability Mitigation. Frost & Sullivan separately ranked Zimperium strongest on both its Innovation Index and Growth Index. The company's sovereign-hosted MTD deployment for the Australian government, launched in November 2023, demonstrated the "Deploy Anywhere" model in practice: local data residency, AGSVA-cleared personnel, a sovereign cloud provider, and IRAP-certified security services eliminated the need for each agency to build and maintain isolated infrastructure. Zimperium has protected U.S. Department of Defense mobile devices for years, a reference it cites when describing the sovereign capability's pedigree.
Microsoft's parallel move centers on ecosystem integration rather than a standalone MTD product. Zimperium lists Microsoft as both a technology partner and a customer — "Trusted by Global Enterprise & Industry Leaders: Microsoft, Trellix, SentinelOne, VMware" appears on the company's site — and the partnership includes integration with Microsoft's security stack. For enterprises already invested in Microsoft Defender for Endpoint and the broader Microsoft 365 security suite, Zimperium's MTD plugs into SIEM, IAM, UEM, and XDR platforms through documented connectors, feeding mobile telemetry into the same dashboards that ingest endpoint and identity signals. This matters because mobile has historically been a blind spot in Zero Trust architectures: MDM solutions manage configuration but lack threat detection, while traditional EDR agents don't run on iOS or Android. Zimperium closes that gap by providing device attestation, app vetting, and network assessment that feed directly into Microsoft's Conditional Access policies and Defender XDR investigations.
The partnership reflects a broader market dynamic: Microsoft secures the PC and identity layers at scale, while Zimperium specializes in the mobile layer with on-device AI that operates independently of cloud connectivity. Neither company is attempting to displace the other; instead, the integration lets security teams extend existing Microsoft investments to cover the mobile attack surface without introducing a second management plane. Zimperium's own messaging underscores the logic: "Unlike solutions that bolt AI on as an afterthought, Zimperium architected AI into the core of its platform from the ground up." Microsoft gains mobile depth without building it; Zimperium gains distribution through the world's largest enterprise security vendor.
Both companies are responding to the same signal: attackers have shifted to a mobile-first strategy, and generative AI is lowering the barrier for creating sophisticated mobile threats. Zimperium's 2026 Global Mobile Threat Report notes that nearly one in twelve employees used a China-based AI tool in the past 30 days, expanding the shadow AI attack surface. The convergence of on-device AI defense, sovereign deployment options, and native integration with the Microsoft security ecosystem gives enterprises a practical path to close the mobile visibility gap, provided they can hire the engineers who understand how to tune and operate these systems.
The Federal Mandate That Changed Procurement
A binding directive issued by the Cybersecurity and Infrastructure Security Agency on June 10, 2026, has given federal agencies a hard deadline to treat mobile devices with the same urgency as servers and workstations, and it explicitly ties that urgency to the accelerating use of AI by adversaries. Binding Operational Directive 26-04, "Prioritizing Security Updates Based on Risk," supersedes BOD 19-02 and BOD 22-01 and extends the Known Exploited Vulnerabilities catalog framework to every asset class reported in the Continuous Diagnostics and Mitigation Federal Dashboard, including mobile devices, cloud assets, printers, and other networked equipment. The directive's scope clause makes clear that "federal information system" covers any system used or operated by an agency or on its behalf that collects, processes, stores, transmits, or maintains agency information, a definition that pulls in third-party and FedRAMP-hosted environments where mobile endpoints are often the least monitored layer.
The directive does not mention a single vendor, but its operational requirements map directly to the capabilities AI-driven mobile threat defense platforms now offer. Phase I, effective immediately, orders agencies to review and update vulnerability management policies. Phase II, due within 60 days, demands updated processes that ingest both the CVE database and the KEV catalog to drive ongoing remediation. Phase III, within 180 days, requires each vulnerability to be remediated no later than the timelines in Table 1 — three days for KEVs with active exploitation and forensic triage, 14 days for other KEVs, and 30 days for non-KEV CVEs above a risk threshold. Agencies must also automate status reporting through the CDM Dashboard and continuously tag all assets reachable from outside the agency network. Mobile devices, which routinely roam across carrier and Wi-Fi networks, satisfy that "reachable from outside" condition by default.
The directive's reasoning section connects the compressed timelines to a specific threat shift: "Cyber threat actors exploit unpatched vulnerabilities, and their use of AI may further narrow the time defenders have to react between patch release and possible exploitation." That sentence, buried in the directive's preamble, is the regulatory signal that manual patch cycles and signature-based mobile device management are no longer compliant. When the same document notes that it "advances priorities for securing federal government networks set forth in Executive Order (EO) Promoting Advanced Artificial Intelligence Innovation and Security and the Cyber Strategy for America," it links the remediation mandate to a broader White House push for AI-augmented defense, not just AI risk management.
CISA has committed to publishing standardized data-schema requirements for machine-level asset tagging within 60 days of the directive's issuance and to conducting annual reassessments of the remediation timelines. The first fiscal-year status report, due to the Secretary of Homeland Security, the OMB Director, and the National Cyber Director, will reveal whether agencies can meet the three-day forensic triage window at scale. If they cannot, the directive authorizes CISA to update implementation guidance and potentially shrink the windows further. For the mobile threat defense market, that regulatory ratchet means the baseline for "compliant" will keep rising, and the AI analytics that turn raw telemetry into triage-ready conclusions will move from differentiator to table stakes.
Where the Money and Talent Are Flowing
Schneider Electric deployed Lookout across 90,000 devices. Henkel extended mobile compliance to Android and iOS across its global workforce. A global smart meter manufacturer used Lookout to secure its migration to Microsoft Intune. These are not pilots; they are production footprints at Fortune-500 scale, and they share a common thread: each customer needed mobile EDR that could correlate fleet telemetry with CVE, CISA KEV, and threat intelligence feeds without requiring source-code access to the apps running on those devices.
Lookout's FedRAMP, SOC 2, and GDPR certifications put it on the shortlist for U.S. defense and civilian agencies. The company's Mobile EDR is the only one with FedRAMP authorization in its category, a credential that matters when procurement officers must justify mobile threat defense to authorizing officials. The same platform that secures Schneider's 90,000 endpoints also generates a vectorized SBOM from application binaries, tracks vendor patch hygiene through a standardized Mean Time to Patch metric, and feeds prioritized risk scores into Continuous Threat Exposure Management platforms, capabilities that map directly to the federal zero-trust architecture requirements updated in 2024.
The operational burden driving these deals is measurable. Organizations now manage more than 2,000 assets per employee on average, deploy 83 security tools from 29 vendors, and staff six to eight specialized security teams. Cyber assets have grown 133 percent year over year. Reported vulnerabilities have risen more than 65 percent annually since 2016, and exploited CVEs jumped 20 percent in the last 12 months alone. Gartner found that four out of five senior risk and assurance executives named AI-driven attacks their top business risk for 2024. Password attacks alone surged from 579 per second in 2021 to more than 7,000 per second in 2024.
Against that backdrop, organizations that fully deploy security AI and automation have cut their average breach lifecycle from 322 days to 214, a 34 percent efficiency gain. The remaining gap is stark: 32 percent of critical vulnerabilities stay exposed beyond 180 days, and Tenable research shows 97 percent of attacks still target known, unpatched vulnerabilities.
Hiring signals reflect the same pressure, with demand growing for roles blending mobile reverse engineering, AI/ML, and cloud-native SIEM integration. Defense contractors and system integrators are recruiting for the same skill set: engineers who can operationalize mobile threat intelligence, build detection logic for on-device ML models, and integrate MTD telemetry into SIEM and SOAR workflows. The convergence of FedRAMP-authorized mobile EDR, AI-driven exposure management, and a vulnerability environment that compounds at 65 percent a year has turned mobile threat defense from a compliance checkbox into a budget line item with dedicated headcount.
What Comes Next: Consolidation and Speed
The mobile threat defense market is moving from point-product purchases to platform consolidation. Lookout's MSEC platform now correlates that telemetry, and threat databases to produce prioritized risk metrics, a capability that positions it as a Continuous Threat Exposure Management (CTEM) node rather than a standalone mobile tool. Zimperium's research on mobile banking trojans demonstrates that malware-as-a-service now targets mobile banking and crypto apps at scale. The exploit timeline has collapsed: AI-generated payloads, automated reconnaissance, and rapid variant production mean signature-based defense is obsolete. The only viable response is behavioral analytics trained on massive mobile telemetry datasets: Lookout claims over 220 million devices in its cloud; Zimperium's cloud analyzes billions of app interactions.
Regulatory pressure accelerates procurement. Binding Operational Directive 26-04 forces agencies to demonstrate continuous monitoring of managed and unmanaged devices. Lookout's FedRAMP authorization and Microsoft's sovereign cloud commitments make both viable for federal contracts. Zimperium's purpose-built mobile architecture — covering both devices and applications — aligns with the directive's emphasis on application-layer visibility. Compliance is no longer a checklist; it requires automated policy adjustment based on vulnerability severity and patch availability, exactly what the AI-first platforms now deliver.
The AI layer itself is evolving. Lookout's AI Visibility & Governance module tracks shadow AI and LLM usage across the mobile fleet, a response to data exfiltration through consumer AI apps. As previously noted, the exploit timeline has collapsed, rendering signature-based defense obsolete; the required response is the behavioral-analytics method described above, with Lookout claiming over 220 million devices and Zimperium analyzing billions of app interactions.
Market structure suggests consolidation. Zimperium took strategic investment from Liberty Strategic Capital and SoftBank; Microsoft builds rather than buys but has a history of tuck-in acquisitions for specialized telemetry. An IPO window depends on recurring revenue visibility; the shift to platform subscriptions tied to CTEM workflows improves that metric. Acquisition logic favors the vendor with the richest mobile-to-cloud correlation engine. Lookout's SBOM-from-binary capability and MTTP publisher scoring create a data moat; Zimperium's on-device dynamic analysis and Microsoft's identity graph (Entra ID) each offer different moats.
The longer-term trajectory is clear. Mobile becomes a first-class citizen in exposure management, not an afterthought. The winners will be platforms that turn fleet telemetry into prioritized remediation across device, app, and identity layers — automatically, continuously, and at machine speed. The malware campaigns are not anomalies; they are the new baseline. Enterprises that treat mobile as a checkbox will breach. Those that embed mobile intelligence into CTEM will survive.
Working in frontier tech? Zero G Talent tracks the openings: see every open ASML role, browse frontier tech jobs, openings at Stripe, and the people building the field.