GenAI framework, Security: separate traditional software security from novel threats to models; adapt security-by-design across development, evaluation, operations and maintenance for natural-language inputs and probabilistic behaviour; deploy input moderation filters tuned to domain risks, explore forensic tools to find malicious code hidden in models, and use threat knowledge bases such as MITRE ATLAS in risk assessment and threat modelling.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.