How do we keep people safe when machines start acting on our behalf?

At the Digital Trust Hui Taumata 2025 in Wellington, a panel of futurists, technologists, and educators tackled one of the most urgent challenges in tech today: how to build trust in automated systems. Moderated by James Monaghan, the kōrero brought together:

• Nicolas Fourrier (Technology Entrepreneur)
• Silona Bonewald (President, LeadingBit Solutions)
• Dr. Michelle Dickinson (Founder, Matter Workshop & Nanogirl Labs)
• Antoni Tisot (Founder, ELab AI)

Together, they explored machine identity, agentic AI, and how Aotearoa can prepare for a future where machines don’t just process data—they make decisions.

Machine Identity Is the New Frontier for Digital Trust

As AI becomes embedded across workflows, organisations are now managing not just human users—but machine agents acting with delegated authority. Machine identity isn’t just about API keys or digital certificates anymore. It’s about context, traceability, and verifiable authority.

“We’re moving to a new organisational paradigm… where all divisions, all functions, all workflows are augmented with AIs.” – Antoni Tisot, ELab

Without clear identity frameworks, these agents could act unpredictably—or even dangerously. Think of them as digital interns: powerful, fast, but lacking real-world context unless guided properly.

Without Traceability and Boundaries, AI Can Go Rogue

The panel didn’t shy away from risks. From hallucinated outputs to dangerous omissions in mission-critical scenarios, agentic AI can—and already does—go off-script. The only solution: strong traceability and clearly defined boundaries.

“If you do not know exactly what those agents are doing, when and where and why… you can easily have things get away from you.” – Silona Bonewald, LeadingBit Solutions

Silona shared real-world examples where AI agents exposed private data or bypassed critical safety checks. Without robust monitoring and identity safeguards, the cost of automation could outweigh the benefits.|

Public Education Is a Missing Piece of the Puzzle

Dr. Michelle Dickinson emphasised a point often overlooked in AI strategy: education. While organisations race to adopt AI, many in our communities don’t even have access to reliable internet—let alone understand what agentic AI means.

“Until we are thinking about human-centred design fit for purpose for our people, we can get excited about all the tech… but I do feel like if we don’t have a strategy around bringing people with us, they’re not going to build that trust in the system” – Dr. Michelle Dickinson, Nanogirl Labs

Michelle has launched a new initiative, Smarter With, aimed at lifting AI literacy across Aotearoa—starting with libraries, retirement homes, iwi, and marginalised communities. Because trust starts with understanding.

Action and Experimentation Will Drive Safer Innovation

While some called for regulation, others pushed for urgent action. If we wait too long to act, they warned, shadow AI will continue to grow unchecked—posing risks to data, safety, and trust.

“Actions are going to educate us. If we start to play with these tools, we’ll understand what’s possible and what’s not.” – Nicolas Fourrier, Technology Entrepreneur

The message: start small, test often, fail fast, and iterate. Only by engaging with these tools directly can people and organisations learn how to use them safely and effectively.

Stand-Out Quotes

“Humans are inherently lazy—if there’s a button that says ‘I’ll do this for you’, we’ll press it.” – Dr. Michelle Dickinson, Founder Matter Workshop & Nanogirl Labs

“Without verifications, you’re going to have problems.” – Silona Bonewald, President, LeadingBit Solutions

“AI is already being used—valid or shadow, it’s happening.” – Antoni Tisot, Founder, ELab AI

What’s Next?

Stay connected with the community shaping this future by joining Digital Identity NZ (DINZ). Whether you’re a policymaker, technologist, or community leader, now’s the time to be part of the conversation—and help ensure AI works for all of us.