To that end, a joint guidance entitled, “Careful Adoption of Agentic Artificial Intelligence (AI) Services” released from the Cybersecurity and Infrastructure Security Agency (CSIA) along with Australian Signals Directorate’s Australian Cyber Security Centre (ASD’s ACSC), National Security Agency (NSA), Canadian Centre for Cyber Security, New Zealand National Cyber Security Centre (NCSC-NZ), and United Kingdom National Cyber Security Centre (NCSC-UK).
Additionally, like other AI services, agentic AI can end up misused or misappropriated, leading to productivity losses, service disruption, privacy breaches or cybersecurity incidents.
Organizations must therefore anticipate what could go wrong, assess how agentic AI risk scenarios might affect operations and establish ongoing visibility and assurance to maintain confidence in their agentic AI investments.
Where possible, organizations should also consider a full spectrum of solutions for repetitive tasks, including reducing or eliminating low-value processes, which may be lower risk compared to agentic AI solutions.
Agentic AI systems consist of one or more agents that fundamentally rely on an AI model, such as a large language model (LLM), to interpret and reason about the state of the world, make decisions and take actions.
LLM-based agentic AI systems contain the LLM itself, alongside external tools, external data sources, memory and planning workflows.
Agentic AI system diagram.
Source: Australian Cyber Security Centre
These components enable the system to perceive its environment and, where applicable, take action to achieve its goals. Compared with traditional LLM systems, agentic AI systems distinguish themselves by accomplishing underspecified objectives, acting autonomously, following goal-directed behaviors and creating long-term plans.
Recommendations
The guidance provides actionable recommendations to help organizations anticipate, assess, and mitigate agentic AI-specific risks. Recommendations include:
- Aligning agentic AI risk management with existing cybersecurity frameworks and risk posture
- Limiting agent autonomy by ensuring agents do not grant broad or unrestricted access – especially to sensitive data or critical systems
- Implementing layered defense strategies, strong identity management, and robust oversight mechanisms
- Conducting comprehensive threat modeling, continuous monitoring, and regular security assessments to defend against evolving threats
Click here for more information on the guidance.