AI literacy is no longer a key skill, it’s a compliance issue

Guest Post by Crystel Robbins Rynne, CEO, HRLocker

Across most organisations, AI adoption hasn’t followed a strategic plan. It’s arrived through day-to-day work. A manager uses ChatGPT to screen CVs, sales teams rely on Copilot to clear admin, finance prototypes a bonus model with Claude, an operations lead asks Gemini to rewrite a sensitive letter.

None of these actions looks like formal AI deployment, yet they all carry regulatory implications. The Central Statistics Office reports that text generation tools are now embedded in routine business processes. That means AI is already influencing decisions about people, performance and money, often without any governance structure around it.

This isn’t an IT issue. IT can explain how a model behaves. It cannot determine whether its output should influence a hiring decision or a disciplinary process. Once AI touches people?related decisions, the responsibility shifts to HR, compliance and governance.

The regulatory trigger point

Ireland’s AI Office opened on 2 August. The following day, Article 4 of the EU AI Act came into force, activating supervision and enforcement of AI literacy across the Union. Many organisations still believe the meaningful obligations sit years away.

The complex requirements for high?risk systems, including recruitment and dismissal, have indeed been delayed until December 2027. But AI literacy wasn’t delayed. It’s the first requirement under active oversight. And it applies to every organisation using AI tools, regardless of scale or sophistication.

You’re not expected to understand model architecture or training data. But once AI is used inside your organisation, you must ensure that staff understand the risks, limitations and boundaries of those tools. That obligation extends to contractors and agency workers. And the level of literacy required depends on the impact of the task. Editing a social post is low?risk. Ranking job applicants is not.

AI, in regulatory terms, includes the everyday tools already in circulation. Free accounts. Built?in features. Automated summaries. Meeting transcription tools. If it’s used for work within the organisation, it is in scope.

How governance failures emerge

The Workplace Relations Commission has indicated that enforcement will be incident?driven. Inspectors are unlikely to appear without a trigger. That trigger will be a complaint, a challenge or an adjudication where AI influenced the outcome.

Until now, disclosing that AI assisted a decision was neutral. Under Article 4, it becomes a point of scrutiny. The focus shifts to three questions: who used the tool, what they understood about it and whether the output was verified. If the answer is that a manager accepted an AI?generated result without checking it, the organisation’s governance position is weakened.

The organisational maturity gap

The CSO reports that AI use is already widespread. One in five Irish businesses uses AI, rising to nearly six in ten among large employers. Yet CIPD Ireland’s latest HR Practices in Ireland research shows that most organisations have no AI guidance, minimal training and leadership teams that do not feel prepared for an AI?enabled environment.

AI adoption has outpaced governance. The tools arrived. The controls did not.

The minimum viable control framework

The European Commission has been clear about what organisations need to demonstrate. There is no certification requirement. No literacy exam. The compliance baseline rests on three controls.

1 – A boundary?setting policy. A short, operational document that identifies approved tools, prohibited tools and information that must never be entered into any AI system. For most organisations, this includes employee data, candidate information and commercially sensitive material. It should also specify where human oversight is mandatory and who is responsible for escalation.

2 – Training that builds risk awareness. Employees must understand how AI can fail. Fabricated facts. Non?existent sources. Biased patterns. Over?confident outputs. Training should use real examples from your organisation, not generic scenarios. Staff involved in hiring, performance or pay require deeper, role?specific sessions.

3 – Evidence of compliance. A record of who was trained, when, on what and which version of the policy they acknowledged. This mirrors existing disciplines in health and safety and data protection. Most HR systems already support this level of record?keeping.

If AI touches recruitment, performance or pay, those areas should be prioritised. They carry the highest risk and will be the first areas examined by regulators.

The strategic upside

AI literacy isn’t just a regulatory obligation. It’s a control that improves decision quality. Staff who understand the limits of AI make fewer errors, rely less on unverified outputs and extract more value from the tools already available to them. Organisations gain cleaner audit trails and fewer surprises when decisions are challenged.

CIPD research shows strong appetite for AI learning among HR professionals. The barrier is not interest. It is structure.

A practical starting point is simple. List every AI tool currently in use, including the unofficial ones. Then ask two questions: who has been trained to use it, and where is that documented?

The second question reveals the governance gap. Closing it is now a compliance requirement. And solving it should be a top priority.

About Crystel Robbins Rynne

Crystel has been with HRLocker since 2013 and now leads the company as CEO. She drives the organisation’s strategic direction, aligning product, operations, and customer experience to support HRLocker’s next stage of growth. Crystel works closely with internal teams and external partners to ensure the company is positioned to meet the evolving needs of employers across Ireland and beyond.

A long?standing advocate for better employee experience, Crystel is recognised as a leading voice in the HR space and regularly shares insights through thought leadership articles, HRLocker’s webinars and industry events. With more than two decades of HR experience, she brings a rare combination of board?level perspective and first?hand understanding of the pressures HR teams face every day. Crystel is the driving force behind a new nationwide online community, the HRLocker Hub, built to connect and support anyone working in HR in Ireland.

About HRLocker

Ireland’s leading people management platform, HRLocker, helps SMEs replace manual HR tasks with simple, secure and intuitive tools. Core HR provides one central location to store, manage and organise all employee data. Perform supports structured employee development. Recruit streamlines hiring. Onboard delivers a smooth start for every new employee. Each product works independently or as part of the wider HRLocker suite. Together they provide accurate and compliant record-keeping that strengthens confidence, reduces risk and supports audit readiness. HRLocker is also home to the HRLocker Hub, a free community where HR people across Ireland learn from and support each other. Find out more at hrlocker.com.

See more stories here.