April 14, 2026

Later this week, I'll be heading to New York for a YPO Corporate Governance Symposium. It's an event I've been genuinely looking forward to: an intimate gathering of leaders wrestling with governance under real pressure.

The agenda includes a full session on AI governance and emerging technologies, and I'm going in ready to listen, learn, and participate.

Most governance conversations I've been part of frame the challenge as: "How do boards oversee AI risk?"

I'm bringing a different question...

A recent Accenture and Wharton report captured it plainly:

"Intelligence may be scalable, but accountability is not."

That sentence deserves more attention than it's getting.

We are in an interesting moment. AI agent deployments are accelerating rapidly:

80% of Fortune 500 companies now have active AI agents running in their operations
Gartner projects 40% of enterprise applications will include task-specific AI agents by end of 2026
McKinsey reports that only 1 in 3 organizations has reached mature AI governance

There's a gap forming. And it's not small.

The Governance Question Boards Aren't Asking

Most boards are wrestling with: "How do we oversee AI?"

That's the right question. For a system that recommends.

But as I've outlined in this series, we are moving from Systems of Record to Systems of Action. AI is no longer just recommending decisions. It's executing them.

Here's what makes this harder than it looks. Two structural forces are working against traditional oversight at the same time.

First: the velocity problem. AI capabilities advance in months. Governance and regulatory cycles take years. By the time a framework is ratified, the system it governs has already evolved past it.

Second: the explainability problem. As AI systems become more capable, they frequently become less explainable. Board oversight has historically depended on understanding causation. That assumption is breaking down.

Which means the governance question changes:

"How do you govern autonomous AI systems - not just oversee AI risk, but structure governance for execution layers where AI is making decisions, not just recommending them?"

"Most boards are structured for human-in-the-loop oversight. That model breaks when the system is the loop."

This is what I'm bringing to the table at the symposium. Not as a risk management concern. As a structural design challenge.

Shadow AI Is Making This Urgent

There's a compounding factor that isn't getting enough boardroom attention.

Approximately 75% of knowledge workers are now using AI tools through unsanctioned, bring-your-own channels. Researchers call it "shadow AI." It's not rogue behavior. It's productivity-seeking behavior. People are using the tools that work.

But the governance implication is serious:

Organizations thought they were deploying AI
Instead, AI deployed itself
Accountability gaps are forming faster than governance frameworks can close them

This is not a technology problem. It's an institutional design problem.

Reframing: Governance as Execution Infrastructure

Here's the reframe I'd offer, and what I'm working through in my own investing and advisory work.

Governance for autonomous AI is not about monitoring risk after the fact. It's about designing the execution infrastructure that makes trust possible in the first place:

Permissioning: What can the system authorize autonomously? What requires a human
Audit trails: Every action logged, explainable, and attributable
Escalation architecture: When does the system pause and defer to human judgment?
Accountability mapping: When an autonomous system errs, who is responsible?

That last question is not hypothetical. Consider AI-driven diagnostics in healthcare: a system that is more accurate than any individual physician but less explainable than any individual physician. The outcome improves. The accountability chain breaks. Someone has to design the governance layer that holds both truths at once.

The same pattern repeats across financial services, logistics, legal review. The domain changes. The structural problem doesn't.

This is not a compliance checklist.

This is a product category.

(I'll note, with appropriate restraint, that I'm actively involved with a company operating at exactly this intersection: Identity, Access, and Permissioning as the foundational governance layer for what autonomous systems can do on behalf of a Financial enterprise (in this case: Credit Unions and Community Banks). The opportunity feels early and significant.)

The organizations and vendors who build this well will have a structural advantage that is genuinely difficult to replicate.

The question I'll be reflecting on at the symposium:

The history of governance failures, from Enron to the 2008 financial crisis, teaches one consistent lesson: the structure was usually fine. The social dynamics were broken. Directors didn't feel safe challenging what the system was telling them.

As AI systems become the executors of board decisions, not just the tools that inform them, the same risk reappears in a new form. The board that can challenge an AI system's outputs, that treats dissent as an obligation and not a liability, will govern better than the board that defers to the algorithm.

When the AI is the executor, what does accountable governance actually look like?

Curious to hear from Board Members, Governance practitioners, and those building governance infrastructure for agentic systems.

Reach me at arif@faris-capital.com

This thesis was originally published on LinkedIn. Join the discussion, add your thoughts, and follow for regular updates.

View on LinkedIn