Kaspersky reports 33,300 attacks using fake AI tools against SMEs, five times more than in 2025

Kaspersky Blocks Over 33,300 Attacks on SMEs by Malware Masquerading as AI Tools

Kaspersky has reported that it thwarted more than 33,300 attacks targeting small and medium-sized enterprises (SMEs) using malware disguised as artificial intelligence tools between January and April 2026. This figure represents a fivefold increase compared to the previous year. The cybersecurity firm’s annual report on threats facing small to mid-sized organizations highlights the most common impersonations and offers accessible countermeasures. Notably, fake messaging applications continue to be the most prevalent vector for these attacks.

Based on telemetry from its solutions deployed within this sector, Kaspersky’s report reveals that cybercriminals primarily exploit well-known brands familiar to users. In the first four months of the year, ChatGPT was used as bait in 42% of AI-related app attacks, followed by Claude at 24% and DeepSeek at 20%. The growing integration of these tools into the daily operations of small organizations significantly enhances the effectiveness of these deceptions.

The identified malicious files predominantly consist of variations of Trojan horses, disguised as harmless files to trick users into executing them. Some variants download additional payloads onto compromised systems. Depending on the specific type, these threats can steal, delete, block, modify, or copy data, making them particularly alarming for small organizations due to their versatility.

Fake Messaging Apps Still Outweigh AI Impersonations

The same telemetry data puts Kaspersky’s fivefold increase in context. During the same period, Kaspersky blocked nearly 415,000 attacks exploiting fake messaging and videoconferencing applications, including Telegram, WhatsApp, Zoom, and Microsoft Teams. This volume is stable year-over-year and is twelve times greater than the fake AI applications. The hijacking of common communication tools remains the preferred entry point for social engineering attacks.

The rise in fake AI tools reflects the attackers’ rapid adaptation to emerging trends rather than a fundamental shift in the threat landscape. “The threat landscape is constantly evolving, and attackers are quick to exploit new trends. For example, during the first four months of the year, our solutions for SMEs detected hundreds of attacks derived from the burgeoning popularity of OpenClaw, an AI tool that’s gaining traction in 2026,” noted Vasily Kolesnikov, a security expert at Kaspersky.

Understanding the Report’s Telemetry Within Its Scope

It’s important to note that the figures in the report represent blocked attacks within the protected environments of Kaspersky’s SME solutions, rather than a comprehensive measure of the entire market. While the scope limits the absolute volume of data, the comparison maintains a consistent perimeter between early 2025 and 2026, lending real credibility to the observed trend within this segment.

Furthermore, the mention of OpenClaw echoes the guidance issued earlier this spring by CERT-FR, which advises against deploying such autonomous assistants on workstations until their security is validated. Attackers are capitalizing on the excitement around these tools even before their applications and internal frameworks stabilize. As a result, SMEs, often lacking the resources to monitor downloads, are frequently the first victims.

Best Practices: Verify Addresses, Be Wary of Links, and Commit to Ongoing Training

The report outlines a few key recommendations for safeguarding against these threats. It advocates systematically verifying download site addresses, being cautious of links received via email, restricting software installation to official sources, and ensuring up-to-date protective measures. “Attackers continuously refine their tactics to exploit human error, making ongoing cybersecurity training for employees essential, regardless of the size of the organization,” added Rodion Pyanov, product lead at Kaspersky.

Finally, the report emphasizes that the effectiveness of protection is contingent on its alignment with the organization’s actual resources. Very small businesses often lack the time and budget to keep their teams informed, making outsourced supervision through a managed detection and response service a viable solution in the absence of a dedicated team. Matching tools to each organization’s cybersecurity maturity is crucial for effective defense.

For SMEs, the key takeaway from the report lies less in the striking figures and more in understanding the geography of risk. Every day, social engineering attacks are concentrated in common channels, initially targeting messaging and videoconferencing tools, and now increasingly aiming at AI assistants. Defense strategies should start with simple download regulations, anticipating the next trending tool that could be used as bait.

Scroll to Top