Many employees are already using AI for drafting, summarizing, analyzing, and problem-solving, with or without formal guidance. But they may not know what constitutes acceptable AI use, how to verify its output, and who’s accountable when AI is part of the work.
This reality demands more than policy. It demands proactive communication. And responsible AI can’t stay siloed in IT, legal, or compliance. This goes beyond being a technology challenge. It’s also a people, process, and behavior change challenge.
Start with principles, but make them practical
The Organisation for Economic Co-operation and Development (OECD) AI Principles offer a solid framework for trustworthy and ethical AI (transparency, accountability, fairness, privacy, human rights, and human-centered values). But principles alone don’t change behavior.
Employees aren’t asking, “How do I align with global AI governance frameworks?” They’re asking:
- Can I use this tool?
- Can I enter this data?
- Can I trust this output?
- Who needs to review this?
- What do I do if something seems wrong?
Responsible AI starts with clear, confident answers.
Technology teams can select platforms, Legal can define risk, and Compliance can set policy. But trust, training, and real behavior change require communication.
Governance starts with data and accountability
Nick Unger, founder and managing partner of AI consultancy AlterityAI, has seen how quickly some organizations stumble.
“Everything with AI starts with your data,” he says. “Most companies don’t know what data they have, what its value is, or who controls it.”
That’s a real problem. AI makes it easy for employees to access, combine, or repurpose information without fully understanding its sensitivity or implications. And polished, confident-sounding output can lead employees to incorrectly assume answers are accurate.
They may not be.
AI readiness requires data readiness. Problems arise faster if AI is trained on or connected to messy, outdated, redundant, or poorly governed content. As Unger puts it: “AI is really good at giving answers. They just may not be the right ones.”
That’s the message every organization should reinforce: AI supports the work, but humans still own the outcome.
Guardrails should enable safe experimentation
Without clear guidance, employees will explore AI on their own. The goal is not to scare them away, but to give them a safe, structured way to discover it.
This means approved tools, clear data rules, role-specific examples they can actually use, and simple escalation paths when they’re unsure. It also means training people on what AI can’t do, not just what it can.
Good guardrails answer practical questions and don’t slow adoption. They build confidence.
Trust is the real adoption strategy
While AI can certainly boost productivity, it can also make people uneasy, worried that their expertise is being undervalued, their jobs may be at risk, or they’ll be expected to do more with less.
Responsible AI communication can’t only be about efficiency gains. It has to focus on capacity, judgment, and trust.
Unger describes one of the most challenging workforce dynamics: “The people most comfortable using AI are often the least experienced and least qualified to verify the output. The people with the most experience are often the least comfortable using AI. You have to close that gap.”
That’s where HR, communications, and change management can lead:
- Help managers communicate clearly and consistently.
- Develop training relevant to real roles and use cases.
- Bring those with expertise and experience into the conversation.
- Build feedback loops so governance reflects how work actually gets done.
Responsible AI is an ongoing conversation, not a one-time rollout. Communicators must translate principles, policies, and risks into actionable guidance for wise AI use.
Need support building your organization’s AI communication strategy or making sure your content is AI-ready? The O’Keefe Group helps companies move employees from confusion to confidence and from policy to practice. Let’s connect.
