AI security has matured rapidly enough that the field is differentiating into specialist roles. The AI security architect designs how AI security operates — patterns, defences, threat models, integration. The AI security manager runs the day-to-day operation — incident response for AI-related events, vendor risk management for AI providers, monitoring AI systems in production, partnering with AI engineering teams on operational security, reporting to executive stakeholders on AI security posture. The roles overlap but are distinct, and organisations operating AI at scale increasingly need both.
What the AI Security Manager Actually Does
Manages the AI risk register — identifying and tracking AI-specific risks across the organisation's AI portfolio. Operates the AI vendor risk programme — evaluating AI service providers, monitoring their security posture, managing the contractual and operational aspects of the relationships. Owns AI incident response — defining playbooks for AI-related incidents, leading response when incidents occur, conducting post-incident reviews. Partners with AI engineering teams on operational security — secure deployment practices, monitoring integration, control implementation. Reports AI security posture to executive stakeholders — translating technical security into governance language for the board and executive team.
How It Differs From the Architect
The architect designs; the manager runs. The architect produces patterns, threat models, design reviews; the manager operates the resulting capability. The architect is engaged on new initiatives at design time; the manager is engaged continuously across the existing AI estate. The architect needs depth in security architecture and AI; the manager needs depth in operational security management plus AI-specific understanding. Both roles need AI literacy; the architect needs it deeper at the technical design level, the manager needs it broader across operational scenarios.
The Skill Mix
Foundational security operations capability — incident response, vendor risk management, security programme operation. AI-specific understanding — AI risks, AI failure modes, AI control patterns, AI vendor landscape. Governance and reporting capability — translating AI security into executive language, partnering with risk and compliance functions. Stakeholder management across engineering, executive, vendor, and regulatory audiences. The role is mid-to-senior management level rather than entry-level; candidates typically come from security management roles with developed AI literacy, or from AI roles with developed security management capability.
A pattern in early AI security operations: an organisation deploys AI broadly, security incidents start arising, and ad hoc response by the broader security team produces inconsistent outcomes because no one owns the AI-specific operational discipline. The remediation is naming an AI security manager — someone whose accountability for AI security operations is explicit and whose capability is deliberately developed. Without the named owner, AI security stays improvised regardless of how strong the architecture is.
The AAISM and Similar Credentials
The AI security manager body of knowledge — reflected in credentials like AAISM — typically covers AI security frameworks, AI risk management, AI security operations, AI vendor risk, AI incident response, regulatory frameworks affecting AI security, and governance reporting. The credential signals structural understanding; applied experience builds the operational fluency that genuine capability requires. As with other security management credentials, the credential opens doors and the experience determines what happens once through them.
Sizing the Role for Your Organisation
For organisations with limited AI deployment, the AI security manager function can be a part-time responsibility within the broader security management team. For organisations with material AI deployment, a dedicated AI security manager becomes necessary as the operational load and specialist judgement requirements exceed what part-time attention can sustain. For organisations operating AI at scale or in regulated contexts, the AI security management function is typically a small team rather than a single role, with specialisation by AI domain (LLM operations, computer vision, traditional ML) or by activity (vendor risk, incident response, reporting).
Components of the Role That Distinguish It
- AI risk register — distinct from the broader security risk register, with AI-specific categorisation
- AI vendor programme — beyond standard third-party risk, with attention to AI-specific concerns
- AI incident playbooks — covering scenarios traditional IR playbooks do not anticipate
- AI monitoring integration — partnership with engineering on the telemetry the operational programme needs
- AI governance reporting — quarterly updates to executive stakeholders on AI security posture
- AI regulatory tracking — staying current with the evolving regulatory landscape
Why the Role Will Continue to Differentiate
AI security is a permanent discipline now rather than an emerging one. The operational load is growing as AI adoption deepens. The regulatory accountability is sharpening as governance regimes mature. Organisations that staff the AI security management function deliberately are positioned for the operational reality of running AI safely at scale; organisations that do not are running AI on borrowed luck, and the luck does not last indefinitely.