Who should be accountable for AI-driven awareness programmes?

Why This Matters for Security Teams

AI-driven awareness programmes sit at the intersection of security operations, human-risk management, and identity governance, so accountability cannot drift to the automation vendor or the training platform. The team accountable for policy, audience selection, escalation thresholds, and data handling must still own the outcomes, even when simulation delivery and adaptive coaching are automated. That matters because awareness content now touches secrets, identity data, and behavioural telemetry, which makes poor governance a security issue rather than a communications issue. NIST frames this kind of ownership as part of broader governance and risk management in the NIST Cybersecurity Framework 2.0, where accountability must remain explicit and traceable. NHIMG research on the State of Secrets in AppSec also shows how security failures persist when controls are fragmented and nobody can clearly own remediation. In practice, many security teams encounter accountability gaps only after a poor campaign, a privacy complaint, or a compromised workflow has already exposed the weakness.

How It Works in Practice

Operational accountability works best when the programme has a named owner who can approve content logic, privacy boundaries, escalation paths, and measurement criteria. Automation can assist with scheduling, segmentation, scenario variation, and just-in-time coaching, but it should not be treated as the decision-maker. The owner must be able to answer four practical questions: who is in scope, what data is used, how exceptions are handled, and when a simulated event becomes a real incident. That is consistent with the governance emphasis in NIST Cybersecurity Framework 2.0 and with NHIMG guidance reflected in the LLMjacking research, where compromised identities turn AI-related workflows into attack paths. A practical accountability model usually includes:

• A security or identity leader who owns policy and risk acceptance.
• A human-risk or awareness lead who owns programme design and message quality.
• A privacy or legal reviewer for employee telemetry and retention rules.
• An operations owner who validates escalation, exception handling, and reporting.

Best practice is to treat the AI system as a controlled service, not a delegated authority. Its outputs may inform decisions, but they do not replace human approval for sensitive content, disciplinary implications, or incident escalation. These controls tend to break down when the programme spans multiple business units and no single owner can approve data use or override automated recommendations because governance becomes diffused across teams.

Common Variations and Edge Cases

Tighter accountability often increases coordination overhead, requiring organisations to balance speed of experimentation against auditability and legal exposure. That tradeoff is especially visible in global programmes, unionised environments, and highly regulated sectors where the same simulation may have different privacy or employment implications by region. Current guidance suggests that accountability should stay with the function that can explain the risk decision, but there is no universal standard for whether that is the CISO, IAM leader, or human-risk director; the right answer depends on where policy authority already sits. In smaller organisations, one leader may own security awareness, identity, and incident escalation, while in larger ones those duties are split but still must converge under a single accountable executive. NHIMG’s DeepSeek breach research is a useful reminder that exposure often comes from hidden data handling and weak visibility, not just from the model itself. The practical test is simple: if the AI programme produces an unwanted outcome, the accountable owner must be able to explain the decision trail and pause the workflow immediately.

Related resources from NHI Mgmt Group

• Who is accountable when an employee uses an AI tool to trigger harmful access?
• Who is accountable when a sanctioned AI tool causes a data breach?
• How should security teams govern AI-driven detection systems that update themselves?
• What makes agentic AI an NHI governance issue?