pc
nid 92338
Apr 30, 2026
ataylor@vixio.com
Canada
Canadian Centre for Cyber Security
financial services
123% of draft length
draft in Generated
Generated
2023 charsCanada's Cyber Security Centre Releases Joint Guidance on Agentic Artificial Intelligence Adoption
On May 1, 2026, the Canadian Centre for Cyber Security (Cyber Centre) released joint guidance on the secure adoption of agentic artificial intelligence (AI) services in collaboration with international cybersecurity partners, including the Australian Signals Directorate's Australian Cyber Security Centre (ASD's ACSC), the United States' Cybersecurity and Infrastructure Security Agency (CISA), the United States' National Security Agency (NSA), New Zealand's National Cyber Security Centre (NCSC-NZ), and the United Kingdom's National Cyber Security Centre (NCSC-UK).
Agentic AI systems are composed of agents that rely on large language models (LLMs) to autonomously reason, plan, make decisions and take actions without human intervention. While agentic AI systems offer powerful automation benefits and can enhance operational efficiency, their autonomous operation across interconnected tools, data and environments introduces significant security risks. The joint guidance addresses these risks and is intended for organisations considering developing or deploying agentic AI systems.
The guidance outlines security considerations related to LLMs and AI, describes key risks associated with agentic AI, and provides best practices for agentic AI developers, vendors and operators. It recommends implementing layered defence and strict access controls to reduce the likelihood of compromise. The authoring agencies also provided tailored guidance on designing and developing secure agents, deploying agentic AI securely, operating agentic AI securely, and defending against future risks.
Organisations should consult the full joint guidance to understand the security controls necessary to protect national security and critical infrastructure systems from agentic AI-specific risks.
REFERENCES:
Canadian Centre for Cyber Security. "Joint guidance on the careful adoption of agentic artificial intelligence services." Accessed May 1, 2026. https://www.cyber.gc.ca/en/guidance/careful-adoption-agentic-ai-services
Agent draft
Not run yet.
Published
2495 charsCanadian Centre for Cyber Security Releases Joint Guidance on Agentic Artificial Intelligence Adoption
On April 30, 2026, the Canadian Centre for Cyber Security (Cyber Centre), together with its Australian, U.S., New Zealand and U.K. counterparts, released joint cybersecurity guidance on the careful adoption of agentic artificial intelligence (AI) services. Agentic AI systems use large language models (LLMs) to autonomously reason, plan, make decisions and take actions without human intervention. According to the Cyber Centre announcement , while these systems offer powerful automation benefits, their ability to act across interconnected tools and environments introduces significant security risks that organizations must address before deployment.
Key findings and recommendations of the Cyber Centre include:
Agentic AI poses distinct security risks beyond traditional AI or software: The guidance identifies privilege risks (over‑broad access, identity spoofing), behavioral risks (goal misalignment, deception, emergent capabilities), structural risks (orchestration failures, insecure third‑party components, rogue agents), and accountability risks (opaque decision‑making, hallucinations, lack of visibility). These risks can cascade across interconnected agents, making failures difficult to predict, observe and contain.
Organizations should embed agentic AI controls within existing cybersecurity frameworks: The Cyber Centre advises treating AI security as part of established cyber practices (defense in depth, zero trust, continuous monitoring) rather than as a standalone discipline. Strict privilege management, authentication of all agent‑to‑agent and agent‑to‑service calls, and software bill of materials (SBOM) for third‑party components are essential.
Practical best practices cover design, development, deployment and operation: Key recommendations include implementing layered security controls, progressive deployment with limited initial autonomy, human‑in‑the‑loop approval for high‑impact actions, continuous monitoring and logging of agent behavior, and use of red teaming and adversarial testing. Until security practices mature, organizations should restrict agentic AI to low‑risk, non‑sensitive tasks and never grant broad or unrestricted access to critical systems or sensitive data.
The Cyber Centre strongly recommends that any organization considering agentic AI conduct threat modelling, establish clear governance policies, and ensure reversibility and risk containment before deployment. The full guidance is available on the Cyber Centre website.