Nobody Trusted the Robot. But Here's the Thing... They Didn't Trust You Either.

Let me tell you about trust. Specifically, the kind of trust that gets written about in World Economic Forum reports with very alarming subheadings. Misinformation. Deepfakes. AI hallucinations confidently telling your attendees that the conference lunch starts at noon when it starts at eleven-thirty. The robots are coming for your credibility, apparently.

And look, I don't want to dismiss any of that. Deepfake detection technology spending is projected to surge by 40 percent in 2026. That is a real number representing a real anxiety. There are real people out there genuinely worried that a synthetic version of their keynote speaker is going to say something unhinged on social media twenty minutes before the opening address. (Honestly, for some keynote speakers, that might be an improvement. But that's a different article.)

Here's what I've noticed in nearly twenty years of events, having managed everything from multi-day national conferences to the kind of team that needs a full program just to keep track of itself: the trust crisis in most event organisations isn't coming from a deepfake. It's coming from the fourth paragraph of the invitation email. It's coming from the fractured, slightly-too-many-stalls energy of an organisation that hasn't stopped to ask the most basic question available to it, which is: what does this actually feel like for the person on the receiving end?

You know the email. It reads like someone typed "write me an invitation for a corporate conference" into ChatGPT, accepted the first draft without reading it, changed the date and the venue name, and hit send before the guilt could catch up with them. It uses the word "synergies." Halfway through, the AI has subtly pivoted to recommending that attendees "leverage core competencies," which makes no sense for a charity fundraiser but does make you sound like you've recently completed an MBA you're not sure you needed.

Your guests can tell. They might not know why it feels wrong. But they feel it, the same way you can tell when a birthday card has been written in a hurry. The words are technically correct. The sentiment is technically appropriate. And yet. There is a specific emotional flatness to it that no amount of exclamation marks can fix.

This is the actual trust problem. Not the deepfakes. The flatness. The sense that nobody thought carefully about the person receiving this.

The Actual Problem (Hint: It's the Data, and It's Embarrassing)

Here's something I've learned that sounds boring but is, in fact, the most quietly catastrophic operational issue in the industry: your data is almost certainly a mess. Not a dramatic mess. Not a "headlines were written about this" mess. Just a slow, grinding, dignity-eroding mess that is sabotaging everything you're trying to do with technology, and by extension, everything you're trying to do for your guests.

Something as innocuous as the inconsistent recording of a state name is enough to break an AI-driven personalisation system entirely. "Vic" in one field, "VIC" in another, "Victoria" in a third, "victoria" in a fourth because someone was tired and the caps lock key felt very far away that afternoon. You end up with a guest receiving a personalised invitation to a Melbourne breakfast event that calls them a "valued Sydney-based leader." Which is not a great start to a trust-building exercise. That guest is now, at best, mildly confused. At worst, they're forwarding your email to colleagues as a cautionary tale.

The irony is that messy data doesn't just break your tech. It breaks the guest experience in exactly the same way that a sloppy prompt does: it signals that nobody checked, nobody cared quite enough, nobody thought about what this would feel like to receive. The mechanism is different, but the impression on the guest is identical.

Clean data isn't glamorous. Nobody has ever won an industry award for excellent naming conventions, and nobody ever will, and that is a genuine shame. But it is the bedrock on which everything else either stands or quietly collapses. Invest in it before you invest in anything else.

The trust stack, in order of actual importance

• Clean, consistently formatted data (unglamorous but foundational)
• Explicit consent: tell people how you're using their information
• Human review of AI-generated communications before they leave the building
• Designing for genuine guest experience, not visible tech deployment
• Then, and only then, worry about deepfakes

On Cultural Resistance, Or: Why Your Senior Leader's Suspicion Is Costing You

There is a particular dynamic that plays out in communications and events teams that I find both deeply understandable and genuinely maddening. A team member submits a piece of writing. Someone senior reads it and says, with a certain forensic squint: "Did you use AI for this?" The team member says no. The senior leader doesn't quite believe them. A low-trust environment quietly takes root, and now everyone's walking on eggshells about their own writing ability, which is a spectacular waste of everyone's time and feelings.

The situation gets worse when, separately, someone does use AI badly, and the result is poor. These two events fuse together in the organisational memory like a minor trauma. "AI doesn't work" becomes the institutional takeaway, when the real lesson was "an unclear prompt doesn't work." Which has been true of every form of communication since the memo was invented. Bad briefs produce bad outputs. This is not new. This is not the robot's fault.

People need space to experiment. They need permission to try, produce something mediocre, learn, and try again with a better prompt and a clearer understanding of what they're actually asking for. That requires psychological safety from above, and it requires leaders who are willing to distinguish between "this was written carelessly" and "this was written with assistance," because those are genuinely not the same thing and treating them as equivalent helps nobody.

But here's the thing that connects all of this: bad prompts, messy data, and cultural resistance to learning are all symptoms of the same underlying condition. The organisation got so absorbed in the mechanics of output that it stopped thinking clearly about the person at the other end of it. Which brings me to a tent.

The Story I Keep Coming Back To

A few years ago, I was running a stall for a charity organisation at a large outdoor public event. Multiple internal departments all wanted their own presence: their own banner, their own 3x3 marquee, their own little patch of real estate in the exhibitors' zone. The administrative instinct was to give everyone their square of turf and consider the problem solved.

This is, in miniature, exactly what happens when an organisation gets absorbed in its own internal logistics. Everyone advocates for their own piece. Nobody steps back to ask what the whole thing looks like from the outside.

So I went to the festival organisers and negotiated one large tent, big enough to be divided into zones, one per department, but cohesive enough to feel like a single intentional destination. And then, because it was summer, because the event always ran in the heat of the day, because I had actually stopped and thought about what it feels like to stand in the sun for three hours looking at stalls, I designed the space around shade and seating.

That was the whole insight. Shade. Seating. The least innovative, most technology-free, most aggressively obvious solution you could possibly imagine. I am not going to pretend it was a stroke of genius. It was just the result of asking "what does this person actually need right now?" before asking anything else.

The stall had record attendance. Staff had real, unhurried, high-impact conversations with people who'd sat down because they needed a break from the sun and stayed because the interaction was genuinely good. The brand was memorable. None of it required an algorithm, a personalisation engine, or a single piece of technology more sophisticated than a fold-out chair.

No deepfake detector was harmed in the making of this result.

So What Does This Actually Mean for 2026?

The authenticity challenge is not, at its core, a technology problem. It is an attention problem. Every issue covered in this article, from sloppy AI prompts to dirty data to organisational cultures that punish experimentation, traces back to the same root cause: someone stopped thinking carefully about the guest and started thinking about the output instead.

Use automation to free up time, absolutely. Streamline the logistics, systematise the follow-ups, let the machine handle the calendar invites and the dietary requirement confirmations and the seventeen reminder emails that nobody wants to write manually. But then take that recovered time and spend it on the moments a machine genuinely cannot produce: the unexpected gesture, the conversation that goes longer than scheduled, the space designed around what it actually feels like to be there.

Treat consent as a design principle, not a compliance checkbox. Tell people what you know about them and why you know it. It stops feeling like surveillance and starts feeling like care. These are, it turns out, quite different experiences.

And when a team member's writing gets mistaken for AI-generated content, pause before you make it a criminal matter. Maybe it's wonderfully efficient. Maybe it could be warmer. Either way, the conversation worth having is about the work itself, not the tools that may or may not have helped produce it.

The robots are not going to steal your event brand's soul. But an industry-wide fixation on what ChatGPT can do, while everyone quietly stops thinking about the guest standing in the sun with nowhere to sit, absolutely might.

Start with the guest. The rest gets remarkably simple.