When two men allegedly managed to smuggle handguns past security at the Melbourne Cricket Ground during a packed AFL clash between Collingwood and Carlton, it triggered more than just headlines—it triggered an uncomfortable question: How did they get through?
The short answer is that the venue was relying, at least in part, on artificial intelligence. The longer answer is that it wasn’t relying on it enough—or perhaps too much.
In an effort to improve traffic flow into the ground, the MCG had implemented the Evolv Express system, a sleek, AI-powered scanner that promises to detect weapons without requiring fans to empty pockets, raise arms, or endure the familiar wanding and pat-downs. The idea is elegant: keep people safe without turning every match into an airport-style checkpoint.
Except this time, it didn’t work. Or rather, it worked, but allegedly, the people operating it didn’t. According to reports, the system flagged the two men for further inspection. The alert was raised. The opportunity to intervene existed. But the secondary screening—manual, human, essential—wasn’t done properly. The guns got in. The men got in. And everyone else was left wondering what might have happened if they’d intended harm.
So does this mean AI failed? Not quite.
AI, for all its promise, does not exist in a vacuum. It augments human capacity; it doesn’t replace it. The Evolv system did what it was designed to do: it identified a potential threat. But that threat detection is only the beginning. Without a well-drilled, responsive human process to follow it up, you may as well hang a sign that reads “This system is being monitored by nobody.”
But before we rush to judgment or call for a return to full pat-downs and metal detectors at every gate, it’s worth looking at the bigger picture—because the challenges here are more complex than just a bad night at the office.
The MCG has a responsibility to keep its patrons safe. It also has a responsibility to get 80,000 people into the venue in a reasonable time. And it has to do both without sending ticket prices into the stratosphere. That’s the reality.
Traditional security models—bag searches, wanding, pat-downs—are slow. Introduce them en masse at a venue like the MCG and you quickly discover the limits of patience and public infrastructure. Long queues, angry patrons, late entries and potentially dangerous crowd congestion at the gates. Not exactly a win for customer satisfaction or public safety.
The solution? Streamline. Automate. Make it smarter. AI offers a way to maintain a high volume of entries with a low friction experience—assuming, of course, that the human component works in lockstep with the technology. When it doesn’t, as we’ve seen, you get headlines instead of a smooth entry.
There’s another layer to consider: cost. Every extra layer of screening, every additional security staff member, every high-end scanner or screening lane comes at a price. That price doesn’t vanish—it’s passed on, in some form, to the patrons. Raise prices too much, and you price out families, members, and casual fans. Keep them too low, and you’re stuck making hard choices about where to cut corners.
Venue operators are walking a tightrope—balancing public safety, customer satisfaction, and corporate viability, all while trying to ensure that going to a football match doesn’t start to feel like a trip through customs at LAX. It’s not an easy job. And AI, as promising as it is, is not a silver bullet.
So, where to from here?
For starters, AI can and should play a role—but only as part of a broader, integrated system—one that includes rigorous training, proper escalation protocols, accountability, and yes, enough boots on the ground to act when the system says something’s wrong.
Secondly, venue operators need to be transparent—not just about what went wrong, but about what’s feasible. If the public wants fast entry, cheap tickets, and world-class security, they can realistically expect two of the three. Something’s got to give.
Finally, we need a cultural shift in how we view security. Too often, it’s seen as an inconvenience or an afterthought—until something goes wrong. But real security isn’t about reacting to disasters. It’s about creating systems where those disasters never get a chance to happen. That takes planning, investment, and a willingness to trust but verify.
The MCG incident wasn’t a failure of technology. It was a failure of process, of people, of priorities. But it also offers a wake-up call—not just for the security industry, but for the broader public who, understandably, want to feel safe but don’t always want to pay the cost of that safety in time, inconvenience or ticket price.
AI may be the future, but it isn’t magic. Without the right support, both from security but also from the public, it’s just another tool—one more ghost in the machine.
