By Claudia Vaccarone, Strategic Communications Advisor

Recently, a European flower retailer emailed a training reminder to its Italian staff, automatically translated within its internal communications platform. The English original addressed “all operators”. The AI translation read: “Tutti gli operatori.” That is the generic masculine, a convention now widely questioned in gendered languages. And 72% of the team were women.

Nobody intended harm. Nobody checked, either. A better version was easy: “Tutti gli operatori e le operatrici”, or the neutral “Tutto lo staff operativo”. Small words send big signals of belonging and help shape an inclusive culture.

Tools that are not neutral

I have spent the last two years studying what AI can do, and what it should not. As a Gen X woman who loves technology, as a multilingual communicator, and as an inclusion practitioner, I am constantly looking for where bias slips in.

Here is the core challenge: communicators have an increasingly powerful stack of AI-supported tools to draft content, measure reach, and analyse sentiment. These tools represent a tremendous opportunity to increase productivity and accelerate workflows. However, they are not neutral.

Whether we use Copilot, ChatGPT, Perplexity, or Claude, these models have been trained on datasets scraped from the open web. And the open web has a structural bias problem. Consider language alone: more than half of the web is in English, yet only about 20% of the world speaks it. Minority voices appear far less often, causing tools to default to one tone, one culture, and one gender. Left unadjusted, they flatten everyone else.

Moving fast, thinking slowly

AI adoption in business in 2026 is robust. While 87% of senior leaders use generative AI at work (Adecco, 2025), 75% of communication functions remain stuck in ad hoc experimentation. Furthermore, only 36% of internal communications professionals feel equipped with the proper literacy skills to use AI efficiently (Gallagher, 2026).

This gap deserves urgent attention. We are deploying tools we barely understand, at high speed, on our most precious asset: employee and stakeholder trust.

The costs are already real. One internal communications director in financial services shared a painful realisation: engagement among non-native English speakers had fallen for two years running. The team initially blamed managers, but an audit revealed the true culprit was AI-generated newsletters. When messages fail to reflect employees, they rarely complain; they simply disengage. Meanwhile, enterprise surveys demonstrate the compounding risks of unmanaged AI: EY’s 2025 survey of 975 C-suite leaders found biased outputs among the most common AI risks (53%), and nearly all respondents reported financial losses from AI-related risks.

Creating a practical governance structure

Without formal frameworks for their AI stack, communications teams risk pulling unverified historical data or stereotypical models directly into active workforce messaging.

While Europe has regulatory guardrails (such as the EU AI Act, GDPR, and the UK Equality Act), regulators move slowly while technology races ahead. Rather than waiting for external mandates, we need to audit our tool stacks and build safe, transparent governance.

A practical way to begin is through a structured five-step de-biasing framework:

  • Audit every AI tool in your stack.
  • Test outputs for bias by submitting identical scenarios with different demographic markers.
  • Adopt a living Transparency Charter built on core commitments like human oversight and inclusive design.
  • Diversify your data sources and human reviewers.
  • Monitor performance continuously.

To test your current posture, you can start with two simple questions:

  1. Could you clearly explain how your department uses AI to the average employee?
  2. And are you truly comfortable with its impact?

Over to comms leaders

I work in English, French and Italian. I know how much can hide in small words. So I now read every AI-assisted message with one question: whose voice is missing? Try it on one message this week. Tutti gli operatori… e le operatrici.

AI governance does not have to slow down creativity; it protects our integrity and ensures our messages resonate with everyone we aim to reach, internally and beyond.

To help you start the conversation with your team, legal colleagues, and IT, I have put together a free, adaptable AI Transparency Charter template with six proposed commitments to start living by (disclosure, data accountability, human oversight, inclusive design, employee voice, annual review). You can download the free AI Transparency Charter template and explore it.

Adapt it, challenge it, and let me know what you discover: connect with me on LinkedIn to share your thoughts or discuss how your team is navigating AI governance. I’d love to hear your experience!