A CEO recently put the AI decision-making challenge to me plainly: “It’s more about maintaining clarity on what actually matters and when to act. It’s easier to over-rotate or hesitate if you’re not disciplined about how you filter signal from noise.”

That’s what I’m hearing more often when advising Chief Communications Officers (CCOs), CEOs, and board members. AI is changing when leaders have to decide — and the risk attached to getting the timing wrong. More data. But not always greater clarity. That’s today’s leadership challenge.

Traditional playbooks prepare organizations for visible events and the steps needed to manage them. That logic often applies to transactions, transformations, and PE-led value creation: define the event, align the stakeholders, manage the message, and execute the plan.

But risk in an AI-shaped environment often forms earlier, outside traditional models of control. Issues may still be uncertain. Facts may be incomplete. The source may be unclear. Yet the consequences for the business can already be real.

A storm in a coffee cup

Tim Hortons, Canada’s iconic coffee chain, akin to Dunkin’ in the U.S., has been associated for decades with on-ramp jobs for local teens. Recently, the company faced a slow-burn problem over its perceived reliance on employees from overseas.     

Then “Josh,” an ordinary-looking kid, began appearing in TikTok videos outside Tim Hortons stores, coffee cup in hand, complaining that he had been “locked out” of a job. Josh was not real. “He” was an AI-generated persona.

A few dozen such TikToks, none of them viral on their own, together reached nearly one million views, even though the originating account had fewer than 400 followers. The effect came from micro-targeting and reinforcement within narrow online communities.

Like the damage to a home caused by a leak behind a wall, a problem like this one can be hard to spot and costly to correct. Tim Hortons said the videos were frustrating and concerning, according to reports, and that it had difficulty getting them taken down.

Tim Hortons is not alone. Gaming publisher Nexon found that influencer videos it had promoted were also AI-generated personas. The company had amplified a faked narrative it thought it controlled.

The lesson is not that every outside narrative undermines the brand. It is that organizations need a way to know which weak signals are harmless, which are forming risk, and which require leadership action.

A blind spot in AI oversight

A decade ago, many boards treated cybersecurity as a technical problem for the CTO or CIO. That changed when they understood the enterprise-level threat. AI is following a similar path.

In some organizations, AI sits largely with the CTO or CIO. In others, CFOs are taking a stronger oversight role as boards question the return on AI investment.

AI-mediated reputation risk does not sit neatly with one function. Different leaders see the same risk differently. General Counsel may see exposure. CCOs may see credibility. CEOs may see business consequences. Boards may see enterprise risk.

That is why AI cannot be managed only as a technology issue. It is also a decision issue.

For more than two decades, I have advised CEOs, boards, and CCOs on how strategic communications must evolve as business conditions change. AI is now rewriting the logic of how organizations are perceived, how reputations are formed, and how organizations must respond.

The old model worked largely from the inside out, where organizations built messages and targeted audiences to maintain discipline.

The new reality is increasingly outside in, where those same audiences now query AI systems that scrape, recombine, and reinterpret information from many sources, raising the risk of inaccuracy or misrepresentation.

In moments of transformative change, companies and their investors face even greater exposure.

Why this is a systemic risk

AI-mediated perception and reputation are systemic risks because they change how information is created, interpreted, trusted, and acted upon across the business. 

Several dynamics make this harder to manage:     

• Speed. Narratives can move faster than internal alignment. By the time facts are clear, the external narrative may have moved, and stakeholder perception may already be forming.
• Volume. Organizations must process more information, commentary, monitoring output, and content. More information can confuse rather than support better judgment.
• Ambiguity. Early signals may be incomplete, contested, or distorted. Leaders may need to make decisions before they know whether an issue is material.
• Amplification. Small signals can spread quickly through hard-to-spot filter bubbles or when picked up by influencers, employees, activists, media, or automated systems.
• Synthetic content. Persuasive, false, or misleading content makes source credibility and verification more important.
• Lower tolerance for deliberation. Stakeholders may expect organizations to respond before traditional governance processes have finished, reading consideration as hesitation.
• Higher risk of overreaction. A company that moves too quickly can legitimize noise, create legal or operational exposure, or lock the organization into a position before facts are clear.

Organizations need a way to classify emerging signals, define escalation thresholds, clarify ownership, and test whether action is defensible — so the CCO, CEO, or board can move with confidence.

Building a strategic response

Adaptation requires strategic answers before ramping up a tactical response. As a starting point, leadership should rethink five areas that make up the operating system for how the enterprise sees risk forming:     

1. Strategy: From response to simulation
   Move from reactive messaging to testing how stakeholders, media, employees, investors, regulators, and AI systems may interpret emerging issues.
   Leadership question: What capabilities do we need to anticipate how perception may form before the external narrative hardens?
2. Content: Clean the source
   Audit and restructure archives, metadata, and public content so AI systems surface accurate, current, and strategically relevant information.
   Leadership question: Where do AI outputs diverge from our intended narrative, and what could that mean for reputation, trust, or valuation?

3. Capabilities: From producers to sense-makers
   Build AI fluency so communications teams can interpret what the machines surface, what they miss, and what those signals may mean for the business.
   Leadership question: Are we developing leaders who can interpret AI-shaped signals, not simply produce more content?

4. Structure: From silo to network
   Traditional hierarchies struggle to keep pace with real-time reputation dynamics. The CCO must be able to orchestrate across legal, risk, strategy, HR, investor relations, operations, and commercial teams.
   Leadership question: Does the CCO have the authority and relationships needed to manage reputation risk across the enterprise?

5. Policy: Define signals, boundaries, and escalation
   Decide which signals matter, which patterns require attention, who owns the call, and when an issue should move from monitoring to escalation.
   Leadership question: What are our escalation triggers before external narratives define us?

Where leadership must make moves

Boards are paying more attention, but governance is still uneven. A Harvard Law School Forum on Corporate Governance report on cyber and AI oversight disclosures found that nearly half (48%) of Fortune 100 company disclosures cited AI risk as part of the board’s oversight of risk in 2025 — triple the 16% that did one year before. Still, the level of attention varied widely, with some listing it as just one of many risks overseen by the board, and others explaining their oversight in more detail.

Governance must continue to adapt as AI transforms enterprise-level challenges from visible events into weak signals that may be hard to trace and whose origin can be opaque. Boards, instead of parachuting in in the wake of an event, should ensure that teams elevate appropriate signals before problems take hold.

CCOs have, in recent years, been gaining ground in an advisory role to CEOs on dynamics tied to business moves. Because they straddle internal and external perspectives, CCOs are often in the best position to influence governance conversations.

AI is a systemic risk, the same way cybersecurity became one a decade ago. The challenge may always be to filter signals from noise. Yet leaders who fail to govern that signal will find the noise governing them.