HCLTech: How to Ensure AI Compliance and Responsibility

As organisations accelerate their use of artificial intelligence, agentic AI is moving systems from passive response to proactive workflow execution and orchestration.

This capability is driving both opportunity and complexity, particularly around governance, compliance, transparency and the need for responsible deployment at scale.

Here, Heather Domin, Head of Office of Responsible AI and Governance at HCLTech, shares her perspective on agentic AI, compliance and ethical innovation.

How do you define agentic AI and why is it so important today?

When I talk about agentic AI, I describe it as the point where AI moves from responding to inputs to being able to proactively carry out workflows. Agentic systems are often enabled to plan and act toward a defined goal with autonomy.

That shift is significant. It means AI is now capable of orchestrating multiple tasks and adapting dynamically to changing conditions. That capability can transform how organisations operate.

Of course, greater autonomy brings greater responsibility. As AI systems begin to initiate actions rather than simply recommend them, governance becomes central. Responsible AI alignment for things like enabling the right level of accountability and transparency are not optional safeguards, but foundational.

At HCLTech, we often help clients use agentic AI as a powerful force for innovation and help ensure it is bounded by clear objectives and responsible AI principles. The future will not be defined not just by how intelligent our systems are, but also by how well we align that intelligence with our values and ethical principles.

What are the biggest challenges in ensuring AI compliance and responsible use?

One of the biggest challenges I see is that organisations often treat responsible AI requires both strong technical domain expertise and coordination with a broad set of stakeholders. In practice, this often means that connecting all pieces requires executive level oversight in the form of AI Ethics Boards or Committees, as well as technical controls embedded within the AI development process. Many organisations may be capable in one or two areas, such as being able to establish process based controls, but struggle in other areas such as the technical controls and automation.

Another challenge is regulatory frameworks and standards, such as the EU AI Act, ISO 42001 and the NIST AI Risk Management Framework, which are raising the bar. Compliance now requires structured documentation, traceability, risk classification and lifecycle monitoring. That is not something you “add on” at the end. Appropriate training for all stakeholders, raising the AI literacy in organisations, is a key piece of the puzzle to help enable the proper understanding of the new bar through the organisation.

In my experience, the real differentiator is integration and starting early. When governance is embedded into the AI lifecycle from the start, compliance becomes an enabler of innovation rather than a constraint.

How can organisations make AI decision-making more transparent and trustworthy?

Transparency often starts with early design decisions, which can prevent situations where it may be difficult to explain outcomes after the fact.

Organisations can clearly document information such as the AI system’s purpose, data sources, assumptions and limitations. Tools such as model cards and structured decision logs can help.

Trust can also be enabled by AI red teaming, bias and testing, security controls and continuous monitoring. When organisations enable responsible AI by design, they can build confidence in their organisation and AI systems.

When should companies choose on-premises AI over cloud solutions?

The decision is often influenced by the context in which the AI system is operating, including the AI use case, risk-level, compliance requirements and organisational infrastructure preferences.

On-premises AI can be appropriate when data sovereignty, regulatory requirements or intellectual property concerns demand tighter control. In highly regulated industries or government deployments, that level of control can be strategically important.

Cloud solutions, however, offer scalability, speed and cost efficiency that many organisations cannot ignore. In reality, we increasingly see hybrid models where enterprises balance flexibility with governance.

The more important question is governance maturity. Regardless of where AI is deployed, organisations should implement robust access controls, monitoring and audit mechanisms.

What ethical risks should businesses consider with autonomous AI systems?

Autonomous systems can operate quickly and autonomously at scale. That brings both benefits and potential risk.

One key concern is goal misalignment. An agent may optimise for a measurable objective while overlooking broader human or contextual considerations. A technically successful outcome can still be ethically problematic.

We must also consider cascading effects. Small errors in interconnected systems can compound and create larger operational consequences.

The answer is not to slow innovation, but to embed structured guardrails. Clear accountability, human oversight and continuous monitoring are essential.

Which AI trends do you think will have the biggest impact in the next few years?

I see three major trends shaping the next chapter of AI. 

First, agentic systems will move from experimentation into enterprise-scale deployment. These systems will not just generate outputs but will coordinate workflows, manage processes and make multi-step decisions across business functions. That shift will redefine operating models. 

Second, we will see rapid growth in physical AI. AI will increasingly move beyond digital interfaces and into the physical world through robotics, smart infrastructure and edge systems. As AI begins to interact directly with real-world environments, the stakes increase significantly. Safety and real-time governance will become even more critical.

Third, governance will become a competitive differentiator. Organisations that embed accountability, fairness, security, privacy and transparency into their AI strategy will build trust at scale. Regulation is evolving quickly and those who operationalise governance early will move faster and more confidently.

The next few years will not simply be about smarter systems, but about responsible intelligence operating across both digital and physical domains.