What if the biggest risk in data democratisation isn’t exposure, but misalignment?

William McMillan, Former CISO at CIA, CSO at Salesforce challenges leaders to rethink how security, governance and value creation intersect as AI accelerates access to data across the enterprise.

Democratisation requires presence, not permission

Data democratisation only works when leadership stays actively engaged. Expanding access without sustained alignment creates silent risk: teams move quickly, but away from shared intent. This isn’t empowerment, it’s abandonment. William draws a clear distinction between autonomy and absence, arguing that leaders must repeatedly communicate intent, validate understanding, and stay connected as access scales.

Move decisions first, then move data

Centralising data “just in case” is now a liability. Enterprises have crossed a threshold where excessive data movement slows value, multiplies attack surfaces, and obscures accountability. Time‑to‑value deteriorates before insight even begins. William advocates for decision‑led architecture, using human‑guided intelligence layers to test value where data lives, then centralising selectively and deliberately.

Security is an operating model problem

Data and cyber risk emerge from organisational seams, not tools. When CIOs, CTOs, CISOs and data leaders operate sequentially, security becomes a bolt‑on and architecture fractures. AI amplifies these cracks. There is a need for shared accountability, aligned incentives, and joint architectural ownership. This makes security a cultural and operational outcome, not a compliance artifact.

Key takeaways:

• Data democratisation requires leadership presence: Empowerment without alignment increases risk, not value.
• Centralisation should follow decisions, not precede them: Time‑to‑value improves when data movement is intentional.
• Shared accountability underpins secure scale: Integrated leadership and culture matter more than organisational charts.