In February 2026, Singapore’s Infocomm Media Development Authority (IMDA) released the world’s first governance framework specifically for agentic AI — AI systems that can take actions and make decisions with limited human intervention.
It made the news, briefly. Most founders I’ve spoken to since filed it under “something for the IT team.”
That’s understandable, but it’s the wrong call.
Why AI Governance Is an HR Problem
AI has quietly become embedded in how we manage people. The tools most businesses use every day — applicant tracking systems, productivity monitoring platforms, performance management software, engagement surveys — are increasingly AI-powered. They screen candidates, flag underperformers, score engagement, and generate recommendations.
These are HR decisions. They affect people’s livelihoods, careers, and experience of work. And when an AI system is making or influencing those decisions, the accountability doesn’t transfer to the software vendor. It stays with you, the employer.
Singapore’s new governance framework recognises this. It sets out principles around human oversight, accountability, transparency, and fairness for AI systems that take consequential actions. In the context of employment, “consequential actions” includes a lot of things you might not immediately think of as AI governance problems.
The HR Risks Hiding in Plain Sight
Hiring: Many companies use AI-assisted tools for CV screening, scheduling, and initial assessment. These tools learn from historical data, which can embed historical biases. If your tool is systematically filtering out certain candidate profiles, you may have a Workplace Fairness Act problem you don’t know about yet.
Performance management: AI-assisted dashboards that surface productivity scores, attendance patterns, or engagement signals are being used to inform performance conversations. If a decision about an underperforming employee was shaped by an AI-generated insight, your process needs to be able to justify that.
Data: Your employees’ personal data — performance records, medical leave history, communications data from productivity tools — is being processed by third-party AI systems. Under Singapore’s Personal Data Protection Act, you’re responsible for how that data is used, even if a vendor is processing it on your behalf.
Shadow AI: Your employees are using AI tools — ChatGPT, Claude, Gemini — to draft performance reviews, write disciplinary letters, create job descriptions, summarise HR conversations. Without a policy, you have no visibility and no control over what data is leaving your organisation.
What This Series Covers
Over the next few months, I’ll be writing a series of posts on the intersection of AI and people management. Not from a technical angle — there are others better placed to talk about AI architecture and governance frameworks. This series is about the people dimension: the HR compliance obligations, the culture and fairness questions, and the practical steps founders and people leaders can take.
The series will cover: whether your hiring algorithm could be breaking employment law; shadow AI in your workplace; how to write an AI use policy your team will actually follow; AI in performance management and the fairness question; and building an AI-ready culture from the people up.
Where to Start
If you’re a founder or people leader reading this, I’d suggest starting with a simple audit. List the AI tools your HR processes currently rely on — your ATS, any performance or engagement software, your payroll system, and any tools your team uses to draft people-related content.
For each tool, ask: what data does it use? What decisions does it influence? Who in our organisation is accountable for it?
Most companies I speak to can’t fully answer those three questions. That’s not a technology problem. It’s a people governance problem.
And that’s precisely where the HR function — or a fractional Head of People who understands this landscape — becomes essential.
Want to Talk Through Your Situation?
If you’re unsure how AI is embedded in your people processes, or whether your current practices would stand up to scrutiny, get in touch. This is the kind of question I help founders think through before it becomes a problem.
