.png)
AI In Operations Needs Its Own Guardrails
Organizations are moving quickly from experimenting with AI to relying on it for customer service, software development, security operations and business decisions. AI agents can now take actions on their own, call other tools and pass work to other agents. Each of those capabilities introduces risks that traditional security controls were not written for.
SCF CORE AI-Enabled Operations brings together the SCF controls that address those risks. It gives organizations a defined way to govern the AI they use, keep humans in the loop and show customers and regulators that AI-driven operations are under control.
SCF CORE AI-Enabled Operations is part of the CORE initiative, which the SCF Council created to turn the comprehensive SCF catalog into focused, actionable baselines. Its controls are derived from the Common Controls Framework (CCF) and calibrated to the risks of using AI in daily operations, giving organizations a starting point they can tailor to the AI systems and agents they actually run.
Human Oversight
People can monitor, intervene in and stop AI systems and AI agents when needed.
Agent Boundaries
AI agents operate within defined permissions, data access and tool limits.
Accountability
AI actions are logged, traceable and explainable after the fact.
Volunteer-Driven & Creative Commons Licensed
SCF CORE AI-Enabled Operations is developed by volunteer cybersecurity practitioners and released under a Creative Commons license. There is no cost to access or implement it.
Expectations For AI Systems & AI Agents
Examples of what SCF CORE AI-Enabled Operations expects include:
Governance
AI governance, identification of AI-related legal requirements and awareness of the AI systems in use
Risk
Risk profiling of AI systems and designation of high-risk uses
Prohibited Practices
Screening for unacceptable uses, such as manipulative techniques, social scoring, emotion inference and biometric categorization
Human Oversight
Human-in-the-loop review, escalation paths and protection against reviewer overload or manipulation
Production Monitoring
Monitoring for performance changes, anomalies and emergent behavior, including collusion between agents
Incidents
Reporting and root cause analysis of serious AI incidents
Agent Controls
Sandboxing, kill switches, prompt injection defenses, identity and impersonation defenses, and limits on delegation, tool use and self-modification
Testing
Adversarial and red team testing of AI agents and detection of behavioral drift
Traceability
Logging of agent actions and sessions, output verification, explainability and bias detection
Training
AI risk management training for the people who oversee these systems
The complete control set is in the SCF download and is updated with each SCF release.
Organizations Putting AI To Work
SCF CORE AI-Enabled Operations is designed for organizations that deploy AI systems or AI agents in their own operations, whether they build those systems or buy them from a vendor.
AI Agent Adopters
Organizations giving AI agents access to internal systems, data or customer interactions.
Service Providers
Organizations that use AI to deliver services and need to show customers how that AI is governed.
Risk & Compliance Teams
Teams that need a defined baseline for assessing AI use across the business or across vendors.
%20(white).png)