AI Cyber Risks Keep Evolving, Enterprise Defense Advances

This week confirmed that AI is fundamentally changing cyber risk. As AI-powered attacks become more frequent, autonomous systems demonstrate offensive capabilities, and new malware campaigns expand across Latin America, organizations are being forced to prioritize identity security, architectural resilience, and AI-driven defense over traditional protection models.

Ready? This is your Week in Cybersecurity!

One in Four Malicious Breaches Are Now AI-Enabled: IBM

AI is reshaping the economics of cybercrime, making attacks cheaper to launch while significantly increasing the financial impact of data breaches on organizations. According to IBM's 2026 Cost of a Data Breach Report, AI-enabled attacks accounted for one in four malicious breaches worldwide, with organizations facing average losses of US$6 million per incident, roughly US$1 million higher than the global average breach cost of US$4.99 million.

Kaspersky Uncovers StrikeShark Cyberattack Campaign in LATAM

A newly identified cyberattack campaign is targeting organizations across Latin America, Asia, and the European Union with a previously undocumented malware designed to evade detection and establish long-term access to compromised networks. Researchers at Kaspersky say the operation, dubbed StrikeShark, has affected government agencies, diplomatic entities, software development companies, and other organizations, highlighting the continued evolution of advanced cyber threats.

OpenAI Autonomous Attack Marks New Era of AI Threats: Fortinet

Fortinet warns that the recent AI-driven security incident involving OpenAI models and Hugging Face demonstrates that autonomous systems can exploit existing infrastructure weaknesses at machine speed, making architectural resilience, rapid mitigation, and segmentation essential for enterprise cyber defense.

OpenAI AI Breached Four More Online Services

OpenAI has expanded its disclosure regarding the first publicly known autonomous AI cyberattack, confirming that the incident extended beyond Hugging Face and affected multiple publicly available online services. The company acknowledged that the AI models involved in the internal security evaluation located publicly exposed credentials online and used them to access four separate accounts across four different services.

MBN Expert’s

Traditional Patching Is Not Viable in the AI Era: CrowdStrike

CrowdStrike warned that AI is dramatically accelerating cyberattacks across Latin America, cutting average breakout times to 29 minutes as organizations shift from endpoint protection toward cloud and identity security. Marcos Ferreira, VP, Sales Engineering LATAM, CrowdStrike, notes that AI-powered threat intelligence, behavioral detection, and agentic security operations are becoming essential to counter attackers who increasingly rely on legitimate credentials rather than malware.