AI Governance Roles Are Being Invented in Real Time — Here's What's Actually Emerging

Unlike security and compliance, which had decades to settle into recognizable organizational patterns, AI governance is a genuinely new discipline being staffed in real time — and different organizations are placing it in meaningfully different places: inside the CISO's org, inside legal/compliance, inside a dedicated data science or ML platform team, or as a new standalone function entirely.

The pattern worth watching

• Organizations that placed AI governance purely inside engineering tend to build strong technical controls (model evaluation, red-teaming) but weaker regulatory and contractual coverage.
• Organizations that placed it purely inside legal/compliance tend to produce strong policy documents with weaker connection to what the AI systems are actually doing operationally.
• The organizations that seem to be landing it most effectively are building it as a genuinely cross-functional function with a single accountable owner — regardless of which existing department that owner formally reports into.