World Day for Safety and Health at Work: why AI governance matters in HSEQ for safety-critical industries

This year’s World Day for Safety and Health at Work focuses on the growing impact of digitalisation and artificial intelligence on the modern workplace. While AI is transforming how organisations approach Occupational Safety and Health, it also raises an important challenge: how to harness its benefits without compromising safety, accuracy or compliance.

For Health, Safety, Environment and Quality (HSEQ) professionals, this is a practical challenge embedded firmly within management systems, audit processes and regulatory frameworks. It also extends into how training is created, reviewed and delivered, and where robust QA (Quality Assurance) and oversight are essential to maintaining standards.

For Simon Burrows, HSEQ Manager and Data Protection Officer at Mintra, this is part of his day-to-day role. Responsible for maintaining management systems, supporting audits and ensuring compliance, he works across the business to ensure digital tools strengthen, rather than compromise, the integrity of HSEQ.

In this Q&A, he shares how AI is shaping his role, why governance and human oversight remain critical, and what organisations need to consider to keep HSEQ accurate, consistent and fit for purpose in a digital-first world.

As an HSEQ and GDPR professional, how is AI starting to influence your day-to-day role?

AI is already reshaping the HSEQ role. We have a responsibility to ensure it’s used safely across the business, while also using it to improve how we work.

Within my team, we use AI to identify gaps, duplication and inconsistencies in policies and procedures. That feeds directly into clearer, more concise information. At the same time, we are strengthening governance frameworks in line with emerging regulation, including the EU AI Act.

Supported by internal policies, audits and QA processes, this allows us to scale and standardise effectively as we grow, while maintaining the right checks and balances.

Where can AI add value in managing HESQ systems, and how does it impact assurance from an audit perspective?

AI adds real value in structuring information, generating checklists and supporting compliance activities. It can streamline documentation and make it easier to map requirements against standards, which improves audit preparation.

However, its effectiveness depends on the quality of the inputs and still requires human review. AI can highlight patterns or flag potential gaps, but it does not fully understand context or operational risk.

AI can indicate surface-level compliance, but it cannot replace the depth of a good audit. True assurance relies on human judgement, the ability to probe, challenge and explore beyond what is presented to uncover inefficiencies and opportunities for improvement.

How do you ensure AI-supported processes still meet industry-specific standards, such as OPITO or ISO?

Standards like OPITO and ISO are highly prescriptive and non-negotiable. Any AI-supported outputs sit firmly within those established frameworks, supported by rigorous review and senior sign-off.

Alongside this, we stay closely engaged with industry bodies such as the Learning and Performance Institute, which are actively exploring the role of AI in learning. Being part of these conversations ensures we stay aligned with best practice, while also contributing to how AI is applied responsibly in safety-critical sectors.

From a data protection standpoint, what are the key governance considerations around AI, and what does effective human oversight look like?

The focus is on responsible use, ensuring no personal or sensitive data is entered into AI tools, protecting company intellectual property, and maintaining compliance with GDPR and ISO 27001 principles around confidentiality, integrity and availability.

But governance alone is not enough; human oversight is still essential. That means being involved at every stage, from structuring prompts to reviewing outputs and applying quality assurance before anything is used.

Training and culture play a key role here. People need to understand not just how to use AI, but how to question and validate it. That combination of clear governance and active oversight is what keeps outputs accurate, relevant and compliant.

Do any of the standards we check against have an AI policy or criteria that should be met? If not, why not?

At present, none of the standards Mintra is audited against include specific AI policies or defined criteria. That is not surprising, as most standard-setting bodies are still in the early stages of understanding how AI should be governed in practice.

There is likely some hesitation around how far they can or should go in advising on AI. In many cases, if organisations are meeting the required standards, AI is simply another tool to help achieve those outcomes, whether that is improving efficiency, identifying gaps or enhancing delivery.

Do any standards ask whether we have our own AI policy and assess it?

Generally, no, although during a recent ISO 27001 audit focused on information security, we were asked about our AI policy, and it was reviewed with interest. AI introduces new considerations around data use, governance and risk, so it is something a good auditor in this space should be thinking about. Beyond that, it is not something we are currently seeing widely across other standards.

Is Mintra unique in having an AI policy, and what does this mean?

We are not unique, but we are relatively early.

With the EU AI Act coming into force this summer, organisations operating within the EU will need to formalise their approach. UK companies are also likely to follow closely, with future legislation expected to align in a similar way to when GDPR was introduced, and the UK had to enhance Data Protection Laws to achieve the adequacy arrangement with the EU.

In that context, having an AI policy already in place puts Mintra in a strong position. There is a gap between how quickly AI is being adopted and how quickly standards are evolving, and many organisations are still defining what ‘good’ looks like.

Taking a considered, proactive approach allows us to contribute to that evolving conversation, sharing practical insight, engaging with industry, and helping to explore how AI can be applied responsibly in learning and safety-critical environments.

Should organisations in safety-critical industries be declaring where AI is used, particularly in content generation?

Not necessarily in every case.

AI can be used effectively without changing the integrity of the information itself.

Ultimately, it comes back to standards.

Responsibility sits with the Operator or Duty Holder; the organisation is accountable for the safety of operations. They set the precedent for what is acceptable across their projects and supply chain.

If information is not properly verified, whether AI is involved or not, standards can slip. When incidents occur, it is the Operator or Duty Holder who is accountable to regulators such as the HSE.

Looking ahead, how will AI shape the skills and responsibilities of HSEQ professionals?

AI will continue to drive efficiency, particularly in areas like documentation and reporting, but it will also raise expectations.

HSEQ professionals will need strong AI literacy, critical thinking and attention to detail. The ability to validate and challenge AI outputs will become increasingly important.

AI is a powerful tool, but the responsibility still sits with people.

As organisations continue to integrate AI into workplace systems and learning, the questions raised this International Day for Health and Safety at Work become increasingly relevant. How do we embrace innovation while safeguarding critical safety standards? How do we ensure efficiency doesn’t come at the expense of accuracy, compliance or safety?

As Simon’s insights highlight, the answer lies in balance. AI has a clear role to play in improving efficiency and supporting scale, but it must sit within robust governance frameworks, supported by human oversight, structured QA processes and a strong culture of accountability.

Ultimately, the organisations that succeed will be those that treat AI not as a shortcut, but as a tool, one that is carefully governed, continuously reviewed and always aligned to the standards that keep people safe.