EU AI Act: HRPA Explores Practical Steps for Governance Success

HR Policy Global’s expert panel explored critical considerations for AI implementation and compliance obligations for companies with operations or employees in the EU. The panelists also provided valuable guidance for building effective AI governance structures applicable to all employers.

EU AI Act risk-based framework:

• Prohibited AI systems include those using biometric data to perform emotion recognition.

• High-risk AI systems include those used in employment contexts where significant decisions—such as recruitment, termination, or task allocation—are made.

• Limited-risk AI systems require transparency measures, such as informing users that they are interacting with an AI system.

A foundational legal obligation across all AI use cases is to ensure workforce AI literacy—employees must understand the basics of AI systems they may interact with.

The CHRO’s role: As CHRO of an AI company, Joanna Daly emphasized the unique position HR leaders hold in steering AI initiatives across the enterprise—not just within HR.

• Establishing principles and educating leaders: Working with the Legal department, CHROs can help establish principles – what decisions should be made by AI alone, supported by AI or reserved for human decision-making. 

  • Ms. Daly partnered with her legal team to do a road show to educate the leaders who will  lead AI initiatives. “We can’t make everyone an expert,” Daly said, “the purpose is to make sure they know to call HR and Legal when thinking about AI solutions.”

• Leading discussions: CHROs ask “what problem are we trying to solve?” Is AI necessary or better than other available solutions and what safeguards do we need to put in place? Discussions can get technical quickly and HR can reframe the analysis focusing on company principles.

• Execution: On the question of how the company should communicate transparently with employees and navigate discussions with different forms of workers’ representatives in Europe, “It is worth spending time now to talk with works councils about the principles and decision-making before you start rolling things out,” Ms. Daly advised.

Works council strategy: Alan Wild cited an OECD report showing that the EU is implementing AI at less than half the pace of US companies - 90% for US compared to 33% for European companies.

• “The 70-year-old system of works councils was not built to handle the rapid, incremental change of AI. You must bring them with you on the journey of knowledge,” Mr. Wild advised.

Governance development: Oliver Patel provided detailed models on the pillars of enterprise AI governance, the leadership approach needed to ensure successful implementation, and the layers of workforce AI literacy.

• Mr. Patel also provided a governance maturity model from Nascent Governance to Transformative Governance—with the latter enabling business transformation and commercial success.

What’s next for AI: The panel concluded by identifying the next phase of AI discussions.

• Agentic AI: As AI systems become more autonomous, HR must work closely with engineers to determine which tasks can be performed independently and objectively by AI and which ones require human oversight.

• Skills and competencies: While technical roles may demand deep expertise, panelists agreed that non-technical traits—like a growth mindset, the ability to simplify complexity, early adoption, and internal advocacy—are equally vital.