AI-BOMs Introduced to Combat Shadow AI Threats in Enterprises

Severity: Medium (Score: 57.8)

Sources: Theregister

Summary

As AI applications proliferate in enterprise environments, traditional software bills of materials (SBOMs) are inadequate for tracking all components. The introduction of AI-BOMs aims to provide comprehensive visibility into AI assets, including models, datasets, and tools. This new approach addresses the challenges posed by 'shadow AI,' which encompasses unsanctioned tools and applications used by employees. Cisco has open-sourced its AI-BOM tool to help organizations identify AI assets and their interconnections. Additionally, Cisco released a Model Provenance Kit to track AI model origins and similarities. The urgency for organizations to understand their AI environments is underscored by the potential risks associated with unknown AI components. This shift highlights the need for enhanced security measures as enterprises increasingly rely on AI technologies. Key Points: • AI-BOMs are designed to provide visibility into AI assets in enterprise environments. • The rise of shadow AI complicates security, as unsanctioned tools may expose sensitive data. • Cisco's open-source AI-BOM tool helps organizations identify and manage their AI components.

Key Entities

• Supply Chain Attack (attack_type)
• Worm (attack_type)
• Shai-hulud (malware)
• Shai-Hulud Worm (malware)
• T1003 - OS Credential Dumping (mitre_attack)
• T1041 - Exfiltration Over C2 Channel (mitre_attack)
• T1195 - Supply Chain Compromise (mitre_attack)