Kaspersky warns AI-native cyberthreats are rising, calls for AI-powered cybersecurity

2 Sources

Share

Kaspersky's Adrian Hia warns that AI-native cyberthreats are outpacing defenses across APAC. The company detected 500,000 unique malicious files daily in the past year, 7% higher than 2024. Over 15,000 malware samples disguised as AI software emerged, while 52% of severe breaches went undetected for over 90 days.

AI-Native Cyberthreats Outpace Traditional Defenses

Kaspersky's Managing Director for APAC, Adrian Hia, issued a stark warning: attackers backed by Artificial Intelligence are moving faster than organizations can see, understand, and respond

1

2

. The cybersecurity firm detected and blocked 500,000 unique malicious files daily last year, marking a 7% increase compared to 2024. This surge signals that AI-native cyberthreats have fundamentally changed the threat landscape, with attackers exploiting AI to scale operations and automate attacks at speeds that overwhelm conventional security measures.

The cybersecurity visibility gap has become a critical vulnerability across the APAC region. Security teams face growing blindspots across IT/OT systems as speed and connectivity reshape modern enterprises. According to Hia, cybersecurity today is no longer just a race against time but a race against invisibility. Recent cyberattacks across APAC demonstrate that the divide between online and physical realms has completely disappeared. Nichirei Corp suffered a cyberattack that disrupted its logistics network just weeks ago, while India's manufacturing sector has become an APAC hotspot for industrial ransomware, with groups consistently paralyzing factory floors and industrial IT services.

AI Agents Create New Supply Chain Risks

AI agents introduce a new supply chain layer that dramatically expands the attack surface for cyber sabotage and cyberespionage. This year alone, Kaspersky identified over 15,000 malware samples disguised as agentic AI software

1

2

. Because agents dynamically depend on third-party frameworks, APIs, and plugins, a single compromised upstream dependency can cascade across downstream systems. This interconnected vulnerability chain means organizations must look beyond just stopping attacks and ask whether they actually have visibility into what's already happening inside their environments.

The threat landscape now includes AI-assisted attacks that leverage machine learning to evade detection, adapt to defenses, and identify vulnerabilities faster than human analysts can respond. These sophisticated threats require organizations to fundamentally rethink their security posture and invest in AI-powered cybersecurity solutions that can match the speed and sophistication of modern attackers.

Undetected Malicious Activity Persists for Months

A new report from the Kaspersky Compromise Assessment division reveals a disturbing trend: in 31% of incidents analyzed, malicious activity in organizations had been going on for over three months

1

2

. Over half (52%) of high-severity compromises were only discovered after 90 days of going undetected. The oldest incident identified over the last year remained undetected for as long as four years, highlighting the severe consequences of inadequate monitoring and detection capabilities.

These findings point to a broader challenge in security operations. Many organizations have invested heavily in security technologies, but technology alone cannot compensate for gaps in monitoring, detection, and operational readiness. When organizations rely on reactive security practices or lack continuous monitoring, attackers gain valuable time to move laterally, escalate privileges, and compromise critical assets. A mature Security Operations Center shortens that window by providing the visibility, expertise, and operational discipline needed to detect threats before they become major incidents.

HuMachine Intelligence: Combining AI and Human Expertise

Kaspersky has been combining AI and human expertise for the last two decades, a philosophy the company calls HuMachine Intelligence

2

. Since 2004, Kaspersky has built and refined advanced Machine Learning models trained on huge volumes of anonymized global telemetry, collected ethically and responsibly from millions of endpoints worldwide. This reservoir of high-quality data has enabled the company to develop AI systems that are not only safe and accurate but also resilient to evolving threats.

According to Adrian Hia, AI isn't an add-on at Kaspersky but has been embedded across the entire technology stack for the past 20 years, enabling faster detection, smarter automation, and consistent protection

1

. The future of cybersecurity lies in the collaboration between AI and human expertise, not choosing one over the other. This philosophy has guided Kaspersky's innovation for years and will continue to shape how the company protects organizations against the threats of tomorrow.

AI-Driven SOCs and Kaspersky Next Solutions

Companies managing complex IT infrastructures and handling massive data volumes can leverage comprehensive solutions from the Kaspersky Next product line, which provides real-time protection, full threat visibility, and robust EDR/XDR investigation and response capabilities

1

2

. Generative AI models within the platform swiftly transform raw data into structured, actionable intelligence for security teams and decision-makers, enabling them to work more efficiently while reducing manual effort and bridging skill gaps.

Kaspersky also helps organizations restore their visibility through managed security services such as Kaspersky Compromise Assessment, Kaspersky MDR (Managed Detection and Response), and Kaspersky Incident Response, which cover the entire incident management cycle from threat identification to continuous protection and remediation

1

. For organizations looking to build or enhance their security operations, Kaspersky offers SOC Consulting during initial setup or when enhancing existing security operations, providing comprehensive guidance to build robust security operations centers capable of defending against AI-native threats.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved