Bill C-8 compliance checklist for critical infrastructure operators

A practical, step-by-step checklist covering the four core obligations under Canada’s critical infrastructure cybersecurity legislation, including the AI layer most operators have missed.

By Krikor Tengerian · Co-founder, SecuritAI Technologies Ltd. · Updated June 21, 2026

Bill C-8 compliance is not a one-time project, it is an ongoing program. The four core obligations (cybersecurity program, incident reporting, cybersecurity directions, supply chain protection) each require documented evidence, tested processes, and continuous monitoring. This checklist gives you the practical action items for each obligation, including the AI-specific controls that most operators have not yet addressed.

For context on what Bill C-8 is and which sectors it covers, see: What is Bill C-8? Canada’s critical infrastructure cybersecurity law explained. For how SecuritAI supports a Bill C-8 cybersecurity program, see Bill C-8 compliance.

Obligation 1: Cybersecurity program

A cybersecurity program under Bill C-8 must be documented, implemented, and kept current. It is not enough to have policies on paper, regulators expect evidence of operation.

• ☐ Written cybersecurity program document covering scope, roles, and governance
• ☐ Risk assessment covering all critical systems, including AI systems in operations
• ☐ Asset inventory that includes AI models, data pipelines, and AI-facing APIs
• ☐ Access control policies applied to AI systems and the data they process
• ☐ Documented security testing schedule (annual minimum; quarterly recommended)
• ☐ AI-specific adversarial testing, red teaming covering prompt injection, jailbreaks, data exfiltration
• ☐ Program review and update process (triggered by significant changes or new threats)

Obligation 2: Incident detection and reporting

Incident reporting under Bill C-8 is time-sensitive. You cannot build detection and escalation processes after an incident occurs. They must be ready and tested in advance.

• ☐ Incident detection tools covering all critical systems, including AI runtime monitoring
• ☐ Defined escalation path: who decides an incident is reportable, and in what timeframe
• ☐ Documented incident report template aligned to regulator requirements
• ☐ Contact list for relevant regulatory authority maintained and current
• ☐ Tabletop exercise completed in the past 12 months covering an AI-related incident scenario
• ☐ Log retention policy covering AI prompt and response logs (minimum 90 days recommended)

Obligation 3: Cybersecurity directions readiness

The government can issue binding cybersecurity directions to designated operators, specific technical or operational requirements that must be implemented on short notice. Being ready to act quickly is a compliance requirement in itself.

• ☐ Designated point of contact for receiving and acting on government cybersecurity directions
• ☐ Change management process capable of implementing technical controls quickly
• ☐ Vendor relationships in place for rapid security patching across AI and IT systems
• ☐ Documentation practice: record every direction received and how it was implemented

Obligation 4: Supply chain security

Bill C-8 requires operators to assess and address cybersecurity risks originating in their supply chain. For organizations using cloud-hosted AI or third-party AI APIs, this directly includes your AI vendor relationships.

• ☐ Inventory of all third-party vendors with access to critical systems or data
• ☐ AI vendor security assessment: data residency, training data policies, incident notification
• ☐ Contract clauses requiring vendors to notify you of security incidents affecting your systems
• ☐ Annual vendor security review process
• ☐ Documented risk acceptance or mitigation for each high-risk vendor relationship

The AI gap most operators have missed

Every item above that references AI represents a gap in the vast majority of C-8 compliance programs being built right now. Traditional cybersecurity programs were designed before AI became operational in critical infrastructure, they cover networks, endpoints, and applications. They do not cover prompt injection attacks, adversarial model manipulation, AI output filtering, or AI audit logging.

Regulators are beginning to ask these questions explicitly. The organizations that answer them now, with documented evidence, will be in a far stronger position than those that address the AI layer after an incident or a direction.

Not sure where you stand? The free AI Security Readiness Check scores your AI setup in 60 seconds, no signup.

References

1. Public Safety Canada, Royal Assent of Bill C-8 (June 2026)
2. Canadian Centre for Cyber Security, National Cyber Threat Assessment 2025-2026
3. OWASP Top 10 for Large Language Model Applications

← Back to the blog