We are on the cusp of a major shift in how HR teams operate. Artificial Intelligence is here, and HR is one of the first business functions really to feel its impact.
It may start with Chat GPT or CoPilot for writing policies, but it will soon be AI-powered recruitment tools, employee chatbots, performance analytics, and even dismissal decisions (and that may already be the reality for many).
These systems make decisions that affect people’s careers and livelihoods. How they are used will affect trust in the workplace and the employer brand. That’s why HR must act now on AI governance—even if the rest of the organisation isn’t ready.
The problem HR faces
HR departments are adopting AI tools faster than many other parts of the business. Why? Efficiency demands, talent shortages, and suppliers pushing new solutions. AI promises to speed up hiring, reduce administration, and enhance employee experience.
But HR AI tools are also some of the riskiest that a business will face. They have a direct, and very significant, impact upon individuals. The high-risk nature of these AI systems is specifically acknowledged in a range of AI legislation that is being introduced in various jurisdictions. Getting it wrong can lead to serious legal consequences, and unethical, discriminatory decisions being taken. At the same time, many organisations—especially mid-sized and scaling ones—don’t yet have a company-wide AI governance framework. For many, they are on the “too difficult” or “too expensive” pile.
These risks won’t go away, but they can be managed.
So HR leaders are caught in a tough spot. The Chief People Officer or HR Director has to make a choice: implement AI tools without proper oversight, or wait until governance catches up and risk falling behind operationally. Neither option is acceptable.
Why waiting is not an option
Compliance risk: Regulations are coming fast. AI used in hiring, promotions, and dismissal decisions is often classified as “high-risk”. Waiting to act means risking non-compliance.
Ethical and reputational risk: A flawed AI tool can unfairly reject candidates or penalize employees, leading to reputational damage and loss of trust. These issues can become public very quickly.
Strategic risk: Poor AI decisions can lead to bad hires, disengaged staff, and internal friction. Worse, HR may miss the chance to guide the broader AI strategy across the organization. If HR can show how to use AI responsibly, it sets the tone for everyone else.
A Minimum Viable AI Governance model
For many organisations, perfect governance cannot be the goal. What HR needs is a Minimum Viable Governance (MVG) model: fast to implement, lightweight, and effective enough to mitigate risk and establish a defensible position.
Here’s what that looks like:
1. Inventory/data mapping: List every AI tool currently in use or under consideration. Include supplier-led solutions and anything staff may be using informally. You can’t manage what you don’t know exists. Whilst this would ideally be an exercise for across the organisation, our focus is on speed, so start with the HR team. Be expansive in what you regard as AI – if it looks like AI, and sounds like AI, then treat it as AI.
2. Risk assessment: Whilst risk is not the only way to classify AI systems, all good governance frameworks give a significant weighting to risk, so conduct a risk assessment. Remember though that much employment-related AI can have significant consequences for employees, so your assessment is likely to trend towards the higher-risk end of the spectrum.
High risk will include anything that has an influence on decisions associated with hiring, performance management, promotions, discipline or dismissal.
Medium risk tools may include bots, or operational HR systems that assimilate and present data and which one is used to improve efficiency or streamline operations. They will often be assimilating or interpreting confidential personal information, or significant amounts of data, so still have some risk attached.
Low risk systems will be confined to purely administrative systems that may automate or improve productivity but have no direct HR decision-making role, and do not process any employee or candidate data.
Remember – risk must be managed, so you will need to do something with your risk matrix. You will also need to make sure that your data protection impact assessments (DPIAs) are up to date for high risk systems. Much of your data protection compliance framework can be adapted for AI – data protection is your friend.
3. Accountability: Assign an owner for each AI tool. The HR team must own the risk associated with tools they deploy. Define roles: who evaluates, approves, monitors? Ensure that the owner understands their responsibilities. Higher risk will demand greater monitoring and oversight. The assigned team member will need to do their homework, and understand the uses for the tool, its limitations and specific risks. They will need to be prepared to engage with the use that the tools are being put to.
4. Policy: Draft a short AI use policy (3-4 pages) covering:
- Fairness and bias mitigation
- Transparency and explainability
- Data privacy and security
- Human oversight and appeal processes
This doesn’t have to be exhaustive. It just needs to set principles and expectations. It should be owned by HR but available to the business as a whole.
Also, ensure that your privacy information notices and similar transparency documents properly disclose the purposes for which AI is deployed and the systems used.
5. Review cadence: Set a recurring review (quarterly is enough) to:
- Reassess tools
- Monitor outcomes
- Update inventories and policy
- Capture employee feedback
What this looks like in practice
Let’s say you’re the HR Director at a 200-person company. Your team uses an AI-powered screening tool for recruitment and a chatbot for onboarding FAQs. There’s no central AI governance team.
Apply an MVG model:
1. Inventory/mapping: List the two tools, and the purposes for which they are used in the organisation. You may already have that in your record of processing activities (A data protection requirement).
2. Risk: Screening tool = high risk; chatbot = medium. Record this, together with the reasons for the grading. Ensure that you have a DPIA for the screening tool.
3. Accountability: Assign one member of the team as the tool lead for each.
4. Policy: Write a short document outlining how AI tools must align with fairness, privacy, and human oversight principles in your business. Ensure that you refer to the systems in your privacy information notices, and other transparency documents.
5. Review: Schedule quarterly reviews in your regular HR operations meeting.
This won’t be you bulletproof. But you’re ahead of the majority of employers. You’ve created structure, identified risk and started to manage it, and shown leadership in a discipline where many would rather ignore it. It gives you operational experience in managing AI systems, and will help as more AI systems are acquired and developed.
Benefits beyond compliance
Leading on AI governance builds HR’s credibility inside and outside the organisation. It shows employees that they matter. It shows leadership that HR understands the strategic importance of AI, and it gives the rest of the organisation a model to build from.
Instead of waiting for perfection, HR becomes the testing ground for responsible AI—and maybe even the catalyst for company-wide governance.
Start small, start now
You don’t need to be a data scientist or legal expert to lead on AI governance. You just need to start. Create a quick inventory, write a short policy, and assign accountability. Minimum viable governance isn’t second-best. It’s the smart, necessary first move.
Contact us
You can contact Matthew Cole for assistance on any of these issues by e-mailing Matthew at mcole@prettys.co.uk