Trust is often treated as a feeling, but researchers describe it as a judgement people make about a leader, resting on three qualities: ability, whether the leader has the skill and knowledge to lead the change well; benevolence, whether the leader genuinely has people's interests at heart; and integrity, whether the leader is honest and keeps their word (Mayer, Davis and Schoorman, 1995). During automation and AI projects, all three come under strain at once. Leaders are asked to lead something unfamiliar, roles feel exposed, and decisions are often made faster than they can be fully explained.
Change is a natural trust risk because it removes the certainty people previously relied on. Hope Hailey and Jacobs (2023) note that trust is quick to lose and slow to rebuild: small, repeated breaches, a decision reversed without explanation, a risk downplayed, and a promise missed do more damage than one obvious failure. During AI and automation change, this risk is heightened because people cannot always see how a system reaches its outputs, so any gap in explanation is easily filled with suspicion rather than fact.
Transparency is the practical discipline that protects trust through this uncertainty. It means sharing not just decisions, but the reasoning behind them: why this option was chosen over the alternatives, what the change will affect, and what remains genuinely unresolved. Leaders sometimes withhold detail to avoid alarming people, but silence rarely reads as reassurance; it reads as something being hidden. Being specific about what is known, what is being tested and what is still being decided gives people something real to respond to.
AI and automation add a further transparency requirement: explaining the system itself, not just the decision to introduce it. The National Institute of Standards and Technology (2021) sets out four principles for explainable AI: the system should give a reason for its output, that reason should be meaningful to the person receiving it, the explanation should accurately reflect how the system actually works, and the system should be clear about the limits of its own reliability. Applied to leadership, this means being able to tell a team, in plain language, what a tool does, why it was selected, where a person still checks its output, and where it is known to struggle.
The 2026 Edelman Trust Barometer found employers to be the most trusted institutions people deal with, yet a significant credibility gap remains between what leaders are expected to do and what they are seen to deliver (Edelman Trust Institute, 2026). That gap closes through consistent behaviour rather than better messaging alone. A vision loses credibility the moment leaders say one thing and do another (Kotter, 1996), so every update, town hall and one-to-one conversation needs to match what actually happens next.
In practice, this means building routine, two-way channels rather than one-off announcements: inviting questions, answering the difficult ones publicly, and admitting openly when something is not yet decided rather than guessing to appear in control. It means naming the impact on roles directly instead of hiding it behind efficiency language, and it means reporting back on commitments made earlier in the change, so people can see that transparency was not a one-off reassurance but an ongoing standard.
Mistakes will still happen: a rollout slips, an output is wrong, a risk is spotted late. What protects trust here is not avoiding every error, but how quickly and openly a leader responds to it. Naming the mistake before it is discovered, explaining what caused it and what is changing as a result does more to protect trust than a confident silence ever will. People forgive an honest correction far more readily than a discovered cover-up.
Trust built this way compounds. Each honest update, each explained decision and each kept promise makes the next one easier to believe, which is what ultimately allows teams and stakeholders to keep moving through automation and AI-driven change together, rather than being dragged through it.
Action Point
Before your next automation or AI update, write down three things your team or stakeholders do not yet know: what will actually change, why this option was chosen over the alternatives, and what remains genuinely uncertain. Share all three honestly in your next conversation, including what you don’t know yet. Notice whether questions and challenge increase afterwards; that openness, not certainty, is usually the sign that trust is holding.