Overview
Severity: MEDIUM | Affected: US CISA & UK NCSC | Category: policy
In a landmark international collaboration, the US Cybersecurity and Infrastructure Security Agency (CISA) and the UK's National Cyber Security Centre (NCSC) have jointly released the 'Secure AI Framework and Evaluation for Machine Learning' (SAFE-ML). This comprehensive framework provides organizations with a structured approach to integrating security throughout the entire MLOps lifecycle, from data acquisition to model retirement. SAFE-ML introduces a standardized threat modeling methodology adapted for AI/ML risks, such as data poisoning, model evasion, and privacy violations. It also offers actionable controls and best practices for securing data pipelines, hardening training environments, and implementing continuous monitoring for deployed models. The framework is intended to serve as a voluntary baseline for organizations developing AI, especially those in critical infrastructure sectors, to build more resilient and secure systems.