Artificial intelligence governance in HR is moving from a curious experiment to an urgent ethical and regulatory priority, experts have warned. 

According to SHRM’s (Society for Human Resource Management) The State of AI in HR 2026 report, 92 per cent of CHRO’s are introducing AI this year. However, only 17 per cent feel prepared to deal with the ethical and legal risk linked to these tools. 

Speaking at HR technologies yesterday (29 April), Rob Winson, head of HR excellence and people and workplace PMO at the UK Health Security Agency, said AI governance had evolved from being a secondary concern into a core organisational responsibility.

“Responsible governance in HR has moved from interesting to urgent,” he explained, warning that organisations were now managing the “real risk” of protecting workplace trust.

Those risks are already affecting employees. Half (48 per cent) of employees said their trust in leadership declined after the introduction of AI-driven performance monitoring, according to the 2026 Edelman Trust Barometer.

Winson said AI was now influencing hiring, performance reviews and workforce planning. He added: “Emerging EU and UK regulations are also placing greater demand on transparency, bias, management and data quality, while also making employees more skeptical about how their data is being used.” 

Aleksandar Lombard, head of data and AI excellence at Swarovski, said the jewellery maker was putting structured governance in place to meet these requirements. 

Under this system, HR tools are automatically classed as high risk because of the potential ethics and compliance risks. “At Swarovski, every system must be registered, risk assessed and approved before deployment,” he explained.

Sandboxes can also be an effective way for organisations to test AI systems before deployment, according to Winson. This is particularly important in the public sector because it is “generally fairly risk averse”, he added. 

Additionally, organisations can face challenges when applying AI to existing processes without first assessing if those processes are still fit for purpose.

"Stop digitising broken processes,” said Winson. “Too many organisations are taking their legacy policies, role structures and approval-heavy workflows and simply automating them. That just locks in yesterday's assumptions. If you don't change how work is designed, better technology just helps you do the wrong things faster."

Change management also plays an important role during AI adoption, according to Rosalyn Rudran, head of HR systems at The Salvation Army. The charity addressed this by improving staff tech skills and implementing an AI strategy. “It’s then about disseminating that strategy through the workplace and making sure it’s adopted,” she said. 

Lombard added that organisations must also improve their employees’ critical thinking skills to help them to effectively evaluate and challenge AI output. "Without critical thinking, we will not manage the adoption," he said.