AI: The Reputation Signal · Issue 3
The function missing from the AI governance table
Elif Güvençer ·
I started this newsletter on a specific premise — frameworks for enterprise-level AI adoption do exist, but rarely address Communications specifically.
The Communications-specific conversation about AI is almost entirely tactical. Content tools. Monitoring dashboards. Answer Engine Optimisation. Useful starting points. Insufficient as a primary response. And in some cases actively confirming the production identity the function has spent years striving to escape.
I said the harder question — what AI is structurally changing for the function — was not being addressed.
I've spent the last few months working through that question. The result is a framework I've just published: The Two Clocks: A Framework for Communications in the Age of AI.
The framework in short: Communications leaders are running two clocks simultaneously, and most are managing only one.
The immediate clock is outward-facing — monitoring, managing content, and protecting reputation in an AI-mediated world. The structural clock is inward-facing — redesigning what the function needs to become to assume the role the AI era requires of it.
The immediate clock has every incentive driving it. It is where the visible risk is. Where the CEO is watching. The structural clock has no such urgency. Which is why it won't happen unless it is made to happen. For AI adoption to work, Communications leaders need to run both clocks, not just the immediate.
The framework sets out five dimensions: setting strategic intent, redesigning from intent rather than from pressure, the new capability base, signal architecture (governing the coherence of the organisation's reputational signal across the surfaces AI systems draw on), and trust consequence. The first three are the work of the structural clock. The last two are the strategic agenda of the immediate clock.
The full framework is here.
This edition picks up where the last one left off and develops the fifth dimension of the framework — trust consequence — where the governance argument lives.
The governance gap AI amplifies
The total picture AI agents draw is not owned by any single function. Each function does its job, on its surface, to its standards. Nobody reads the whole.
That was manageable before. AI changes the calculus. There is now an intermediary capable of reading the sum of all parts and handing what it finds to everyone who asks.
Inconsistency was always there. It was never this visible, and its cost never this high.
The governance response — without Communications in it
The boardroom data on AI risk has shifted fast. In 2023, 12% of S&P 500 companies disclosed a material AI risk in their annual filings. By 2025, that figure was 72%. Reputational risk sits among the most consistently cited categories.
That shift reflects something larger than the mechanics of disclosure. How an organisation deploys AI, what it discloses about that deployment, and the economic, social, and environmental consequences that follow are now reputational variables in their own right.
Public sentiment is one measure of the stakes. Stanford's 2026 AI Index Report found that 73% of AI experts expect the technology to improve how people do their jobs, while only 23% of the public agrees. No responsible AI statement closes a fifty-point gap.
Inside organisations, the response to this exposure is taking organisational shape. AI governance frameworks are being written. The Chief AI Officer (CAIO) role has emerged with a remit that extends well beyond technical governance — into change management and organisational positioning. Around the CAIO sits a recognisable review structure: Legal, IT, Risk, Compliance, with HR now joining the core.
Communications is the function most often missing.
The absence is not neutral either. The contribution Communications makes to AI governance is one no other function can replicate: it sees the trust consequence in decisions that look defensible from every other vantage point.
When the chain breaks
Communications is the function that works backwards from reputational damage. Something happened. Someone made a call. A context shaped the call. A pressure shaped the context. The work is to reconstruct the sequence well enough to respond — to correct, to clarify, to mitigate. The reconstruction is not complete, but it is reliable enough to act on.
AI-mediated systems remove that reliability. When the reputational damage originates within the system — a hallucinated answer, an agent operating outside its brief, synthetic content circulating without provenance — the reputational consequence arrives in the Communications inbox through a process that is difficult to reconstruct, harder to correct, and, in many cases, not yet possible to attribute cleanly.
What Communications uniquely brings to the AI governance table
AI governance is taking shape as a matrix responsibility. The CAIO, Legal, IT, Risk, Compliance, and, increasingly, HR each carry defined roles. None of those roles is structured to deliver the total trust consequence view.
The CAIO mandate sits closest to that view, but the CAIO is also the architect of the AI agenda. What that agenda is producing in reputational terms needs a second reader — someone whose job is not to drive the work but to test what it generates, independently. That is what Communications brings.
Independence is the point. Every other function comes to AI through a particular lens — compliance, model performance, adoption velocity, brand. Communications comes to it through coherence: whether the parts add up to a single picture that the organisation can stand behind.
Communications sits in a unique horizontal position across the organisation. It shares surfaces with Marketing, HR and every function that produces content the organisation puts into the world. No other function has that reach simultaneously. That position makes the second reading of AI deployment possible.
It is also the only function that, from long experience, knows that legal compliance and reputational coherence are not the same discipline. A statement can be technically defensible and reputationally untenable. A disclosure can be complete and trust-eroding. AI governance has not yet absorbed that distinction.
Deploy the system, own the consequence
As agentic AI becomes more prevalent — systems that act autonomously on behalf of the organisation, without human intermediation at each step — the outputs of those systems are the organisation's outputs.
Air Canada's 2024 tribunal ruling is the canonical example. The airline was held liable for refund guidance fabricated by its chatbot. The airline's defence was that the chatbot was a separate entity responsible for its own outputs. The tribunal rejected that entirely: deploy any system, whether rules-based or generative, own what it produces.
The governance muscle Communications builds now — the discipline of contributing the total trust consequence view, the credibility earned through the matrix governance model — is preparation for a harder version of the same question that is already forming. The time to build it is now.
Where this lands
Narrative coherence and the trust consequence view belong in the design of agent deployment, not as after-the-fact additions.
The function in the room when AI systems are being designed will shape the trust consequences. The function that arrives after deployment will manage it.
Communications knows too well which configuration is more costly.