I recently completed the Generative AI for Business program at the Rotman School of Management, University of Toronto, a program that included a section on AI governance stressing the importance of cross-functional approaches to address issues and opportunities.
And yet. The cross-functional aspect seems to be missing.
Where is the communications function?
In October 2025, a survey conducted by Leger on behalf of the Canadian Public Relations Society showed that only 18% of communications and public relations professionals felt they had a role helping their organization with AI adoption and transformation in Canada. A PwC survey published in February 2026 corroborated this: 90% of companies – 9 in 10 – place responsible AI with IT.
The PwC survey didn’t specifically name PR, communications or marketing. The closest category was customer experience, at 25%. Only 16% place responsible AI with ethics.
Resources follow the same pattern: AI budgets flow primarily to Chief Technology Officers (69%) and Chief Information Officers (60%), with Chief Marketing Officers receiving only 9%.
PwC put it plainly:
AI governance isn’t purely a technical challenge — it’s a business risk issue, a legal issue, and increasingly a board-level issue. When responsible AI lives only in IT, it becomes siloed from strategic conversations where tradeoffs are made and resources are allocated. PwC
PR and communications professionals have traditionally been called upon to extinguish fires precisely because they hold a multi-dimensional view of the organization and its publics. Many actually know what causes fires , and how to prevent them. Keeping them out of AI governance decisions means organizations are missing a function function well suited to anticipate reputational, ethical and stakeholder consequences before they escalate.
Then came Mythos.
Mythos is a taste of things to come
On April 7, Anthropic announced Claude Mythos Preview, a new general-purpose AI model that performs well across the board, but whose standout capability is in computer security. According to Anthropic, you don’t even need to be a security expert to use it to find and exploit sophisticated software vulnerabilities.
Experts are divided on exactly how transformative the model is, but they broadly agree it represents a meaningful leap forward. Anthropic itself deemed it too consequential for a public release, a decision so unusual that it hadn’t been made by a major AI developer since OpenAI temporarily withheld GPT-2 in 2019.
Instead, Anthropic launched Project Glasswing: a controlled initiative bringing together AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, NVIDIA and Palo Alto Networks, plus over 40 additional organizations that build or maintain critical software infrastructure. The goal is to use Mythos defensively, scanning and securing systems before bad actors can exploit the same capabilities
The security, ethical, reputational and governance dimensions of that reality extend far beyond IT.
A seat at the table, before the fire starts
Mythos illustrates something that will only become more common: AI decisions that are technical in nature but organizational and societal in consequence. Decisions about data diversity, biased outcomes, who to involve, and how to communicate risk are not IT decisions. They are governance decisions, and communications professionals are uniquely positioned to contribute to them given the cross-functional nature of the function and ability to understand multiple perspectives and help surface blind spots.
However, the surveys suggest that non-IT functions, PR and communications included, are likely not ready for what’s coming. Being ready requires understanding an organization’s AI capabilities, vision and strategy alongside its ethics, biases, reputational exposure, legal environment and stakeholder landscape: all the dimensions that the PwC data shows are currently left out.
To support early involvement, communications professionals can show how their skills, like stakeholder analysis, narrative development, and risk communication, help spot reputational issues before they grow. They should share insights on situations where communication gaps could affect trust or compliance. By starting conversations with IT, legal, and executive teams, and suggesting communication risk assessments during AI planning, communications leaders can show why their role is essential in responsible AI adoption.
