AI driven attacks rise with weaponised configs and credential theft

The AI Threat Landscape Digest for March-April 2026 outlines significant advancements in the use of AI technologies for cyber-attacks. Key points include: 1) AI attacks have shifted from experimental state-sponsored applications to practical criminal exploitation, utilizing commercial AI models like Claude Code in multi-week operations. 2) Malicious configuration files are being weaponized to bypass safety features of AI systems.

3) AI-enabled platforms are commercializing offensive capabilities, making it easier for attackers to access sophisticated tools. 4) Credentials for AI services have become valuable targets, highlighting a trend of large-scale harvesting from compromised systems. 5) Notable incidents such as the Mexico breach demonstrate the integration of AI in cyber espionage and mass exploitation campaigns, underscoring the evolving threat landscape that both attackers and defenders must navigate.