In the first half of 2018 I ran a five city keynote tour, plus a virtual one, talking to mortgage brokers about what was arriving for them. This was before the Royal Commission had reported. My foresights and recommendations were already written and delivered by the time the commission and the media started making their comments, which is the only reason I felt entitled to say afterwards that the advice still stood.

What I told those rooms was fairly simple. Stop thinking about the rate of change and start looking at the range of change surrounding you. Understand the technology, use it, and don’t let it cost you the human part of the work. You’re not loan processors, you’re finance futurists, and your job is to assess and provide for someone’s future with every loan you write. And then the line I’ve repeated more than any other from that year: we’re in a people’s business, and people love getting advice from a person they can see and trust. Keep humanity at the front and centre of everything you do.

Seven months earlier, on ABC Far North, I’d made the other half of the argument without noticing it was the other half. Trust is fragile, I said, and once it’s broken it’s difficult and sometimes impossible to earn back. The old ways we worked out who to trust, the tribe, the elders, the people who vouched for people, don’t survive contact with a world where we’re constantly assessing strangers we’ll never meet. And the new artificial rating systems don’t fix that. Reputation, I said then, was becoming the ultimate personal currency.

I still hold both of those positions. Neither of them needs rescuing.

What I didn’t follow through was the inverse, and the inverse turned up last week.

If wisdom and trust are the most valuable things left in a transaction, then trust is also the most valuable thing in that transaction to steal.

What AUSTRAC actually found

On Tuesday 19 August, AUSTRAC’s Fintel Alliance published the findings of Operation Claw. Ten major Australian lenders pooled their data. The joint analysis identified potentially hundreds of millions of dollars in suspected fraudulent loans, with reporting since pointing at as much as four billion dollars in property purchases. Sydney is the epicentre. Hundreds of mortgage brokers, lawyers, accountants and high risk companies have been referred to police, to tax authorities and to the corporate regulator.

The mechanics are not exotic. Inflated incomes. Misrepresented employment. Fabricated or unverifiable business activity used to prop up an application. Offshore or third party money used to complete settlements and then to make the repayments, so the loan looks serviceable from the outside while the income that’s supposed to service it doesn’t exist.

None of that is the finding that matters.

The finding that matters is a single word in the AUSTRAC material, and the word is recurring. The same mortgage brokers. The same accountants. The same law firms. Appearing again and again across applications that had nothing else in common. That pattern is what took this from a pile of individually dishonest files to what AUSTRAC has called coordinated professional enabler networks.

Read that carefully, because it’s a different crime from the one most of the coverage is describing. What was being supplied was not a fake payslip. Anyone can produce a fake payslip. What was being supplied was the professional standing that makes a payslip believable. The signature at the bottom. The person on the file who is meant to have looked the borrower in the eye.

That’s the product. That’s what had a price on it.

And you can’t build a business selling that unless somebody, somewhere in the chain, has already stopped checking, because they were told they didn’t need to. Which is exactly what a trusted intermediary is for.

The eight years in between

Here’s what happened to that industry in the eight years since.

In the March 2013 quarter, brokers wrote 44.2 per cent of new residential home loans in Australia. In the March 2026 quarter they wrote 81.0 per cent, and settled $124.88 billion in the process.

The industry didn’t survive. It won.

And this is where I’d apply Ripple Effects, because the second order consequence of winning is the one nobody costed. When an intermediary layer goes from being a bit under half the market to being four fifths of it, it stops being a service and becomes infrastructure. And infrastructure gets treated completely differently. Infrastructure gets regulated. It gets audited. It gets a duty written into legislation about whose interests it serves. And, inevitably, it gets attacked, because attacking infrastructure is efficient in a way that attacking a service is not.

None of that happens to a channel carrying 44 per cent. All of it happens to a channel carrying 81 per cent.

The Royal Commission, the Best Interests Duty, one major bank referring itself to the New South Wales Police and to ASIC over around a billion dollars of suspected fraud while AUSTRAC investigates the sector independently, and now Operation Claw. These aren’t a run of bad luck. They’re the ordinary weather of becoming load bearing.

I want to be very clear about something, because I’ll be standing in front of a room of mortgage brokers on 2 September and I’d say it there too. Hundreds of referrals inside an industry of many thousands of practitioners is not an industry of criminals. It’s a small network that found a soft spot. The problem is that the cost of a network finding is never paid by the network. It’s paid by everybody who wasn’t in it, which in this case is nearly everyone.

The word everybody uses, and the thing almost nobody names

Every organisation I walk into talks about trust. It’s on the wall. It’s in the values statement. It’s the third bullet point in the strategy deck, usually between innovation and customer centricity.

Naming trust is the cheapest thing in the room. It costs nothing and commits you to nothing.

Here’s the difficulty, and I don’t think it gets said often enough. Trust is an outcome. It’s a state that exists in somebody else’s head, about you, and you don’t have access to it. You can’t do trust. You can’t put trust on a project plan. Nobody has ever gone into a Monday morning and executed some trust.

What you can do is the specific set of things that produce it. And that’s where most organisations, and most professions, have nothing at all. They’ve named the destination and left the road blank.

That’s not an accident, and it’s the thing I’ve spent the past year working through. Somewhere in the last two hundred and fifty years we separated knowing from doing, so that fewer and fewer people ever hold a whole piece of work. We built a hierarchy of knowing that turned out to be a map of where information was stored rather than of how anybody thinks. We decided wisdom was scarce, when what was actually scarce was permission to be seen as wise. And we settled on the credential as the proxy for judgment, which worked tolerably well right up until the moment it didn’t.

What’s left underneath all of that is the transaction. Fast, cheap, repeatable, and utterly interchangeable. And a transaction model works beautifully until the thing you need is precisely the thing a transaction can’t carry.

I’ve been calling what’s missing ceremony.

Not ceremony in the sense of pomp, or a formal occasion, or anything with a lectern in it. The distinction I’d draw is between ritual and ceremony, and we use those two words interchangeably, including me for years. A ritual is a form you repeat, and its value is in the sameness. That’s not a criticism. A good ritual is worth a great deal, you can train it, schedule it and rely on it, and most of what organisations do well they do as ritual. Ceremony is the other thing. It’s a response you could only have given to this one person, and its entire value is that it couldn’t have been given to anybody else.

Which is also why organisations find it so hard, and I don’t think the reason is that their people are unwilling. It’s that anything whose value is in the sameness can be specified, documented and handed on, and everything we’ve built to run an organisation is designed to capture exactly that. The workflow, the quality assurance, the CRM, the compliance file. All of it can see the part that repeats. Almost none of it can see the part that doesn’t, which means the part that doesn’t leaves no trace, earns no credit and eventually stops being done.

That’s a design problem rather than a character problem, and design problems can be worked on. It’s the only optimistic sentence in this piece and I mean it.

I’m not going to pretend I have evidence that ceremony prevents an organisation from falling off a trust cliff. Nobody has that evidence, including me, and I’d rather say so than dress up a conviction as a finding. What I’ll assert is narrower and I think defensible: trust is the outcome we all keep naming, ceremony is the only mechanism I’ve found that reliably produces it, and an organisation that names the outcome without ever naming the mechanism has written itself a wish, not a strategy.

Where the trust cliff actually sits

Trust Cliffs is my name for the observation that trust rarely erodes gently, and I wrote in July about why a high trust reading is not the safety signal it looks like. It holds, and holds, and holds, and then it goes, and nobody sees the edge until they’re past it.

The cliff in this story is not the fraud. The fraud is the thing that reveals where the cliff was.

The cliff is the moment a lender, a regulator or a borrower stops treating the professional signature on a file as evidence and starts treating it as a claim requiring verification. That shift costs almost nothing to make and is close to impossible to reverse, and it applies to everybody in the category, because a network finding is a finding about the category.

If you want to know how quickly that can happen, look at where you personally sit right now on a phone call from a number you don’t recognise, no matter who the person says they are. That change took about a decade and it is not coming back.

Specific is not the same as true

There’s an objection to all of that, and it deserves an answer rather than a dodge.

Somebody will say a machine can’t do the part I’ve just described.

I’ve never argued that and I’m not about to start. From the moment somebody worked out that a stick and a rock would move something they couldn’t shift on their own, we have been looking for things to do the work for us. The lever, the wheel, the steam engine, and everything in between. The only thing that’s changed is which kind of work. For nearly all of our history we outsourced our brawn. Now we’re outsourcing our brain. That’s a continuation, not a rupture, and it looks like a rupture only if you’re watching the news rather than the arc.

So yes. Give a system enough about a person and what comes back is genuinely specific to them, and the person on the other end genuinely receives it. I’m not going to tell anybody who has had one of those conversations that they didn’t have it. Whether that undoes my argument or finishes it is a longer piece than this one, and I’ll come back to it properly.

What it doesn’t settle is the thing this week has put on the table.

Every one of those loan files was specific. Real names, real circumstances, a professional who knew enough about the borrower to misrepresent them convincingly. Specificity wasn’t in short supply anywhere in Operation Claw. Specificity was the product.

Being specific about somebody is not the same as being true to them. Those two things look identical from the outside, they feel identical to the person receiving them, and almost every system we’ve built to detect one of them measures the other.

I don’t think many organisations have looked into that gap. It’s where a great deal of this is going to be decided.

Three ripples ahead, and not one of them is about AI taking the job

I keep being asked whether AI is going to replace the broker. I think it’s the wrong question, in the same way that asking whether the calculator replaced the accountant was the wrong question. Here are three things moving right now that will do more to reshape this work than any model will.

How the work gets paid for is not a fixed feature of the universe. From 1 June 2026, Westpac in New Zealand scrapped trail commission, and required existing trail books to be sold rather than simply run off. The Netherlands banned commission on these products outright some years ago and the market restructured around advice fees. I’m not predicting either of those for Australia. I’m pointing out that a remuneration model most people treat as permanent is a policy choice, and policy choices are exactly the sort of thing that get made somewhere else and arrive on a Tuesday.

The verification side is automating before the advice side is. Experian reported on 11 August that 72 per cent of institutions are already running agentic AI in underwriting. That’s the lender’s half of the conversation, which is historically the half that absorbed all the paperwork and much of the waiting. When that half gets faster, the value of the other half stops being measured in how efficiently you assemble a file. There is a cost to all this verifying, and I have written before about the checking tax that almost never appears as a line item.

And the reason people borrow is being rewritten underneath everybody. Negative gearing changed at 7:30pm on 12 May 2026. The capital gains change applies to gains accruing after 1 July 2027, reaches individuals, trusts and partnerships, and includes a 30 per cent minimum tax rate on capital gains rather than just an adjustment to cost base indexation. That’s not a housing policy change. That’s a change to the arithmetic behind a large slice of why Australians take on debt at all.

Now hold those three next to one number. Connective found that 86 per cent of the industry believes AI matters, 3 per cent have it embedded in how they work, and 3 per cent have a policy about it.

Eighty six, three and three.

That gap between believing and doing isn’t laziness and it isn’t stupidity. It’s what happens when people can see something arriving and have no idea which part of their work it lands on. That’s not an AI problem. That’s a question about what work is actually yours, which is the whole point of HUMAND, and it’s answerable in an afternoon by anybody willing to write down what they actually do all day.

If you’ve never written a home loan in your life

I’ve used broking because it’s the clearest worked example available this week. But almost none of this is about broking.

Nearly every organisation I’ve ever worked with has an intermediary layer it depends on and cannot actually check. A recruiter who tells you this candidate is good. A certifier who signs off that the work meets the standard. A reseller who represents you in a market you’ve never visited. An agency, an auditor, an accreditor, a franchisee, a contractor’s contractor. Somebody, somewhere in your operation, is exercising judgment on your behalf, and you have accepted their judgment as fact for so long that you’ve forgotten it was ever a decision.

That delegation is invisible while it works. It only becomes visible on the day it doesn’t, at which point everybody discovers simultaneously that nobody had written down who owned that call. Decision Trust Zones is the frame I use for this, and I applied it last week to what happens when the institutions you rely on stop agreeing with each other. Its first question is deliberately uncomfortable: not who does this work, but on whose judgment does this rest, and does that person know it rests on them.

It’s worth adding one more thing, because it changes the maths for everybody. In the first quarter of 2026, across more than nine million identity verification transactions, AI generated document fraud passed physical forgery for the first time. The confirmed fraud rate ran at 3.89 per cent, which is roughly one in every twenty six attempts. Every document your organisation currently accepts as proof of something was designed for a world in which producing a convincing fake was expensive and slow. That world has ended, quietly, without an announcement, and most verification processes have not been told.

So here’s the practical thing, and it takes about twenty minutes.

Write down the last five decisions your organisation made that depended on somebody outside it being honest and competent. For each one, write what you actually relied on. Not what the policy says. What you actually relied on. Then ask, for each one, what you would have to see to know it was wrong, and how long it would take you to see it.

If the answer to that last question is longer than a year, you don’t have a control. You have a habit.

What I’d say to the room, and what I’d say to you

I’m in front of a room of mortgage brokers on Wednesday 2 September, carrying this same conversation on with them in person, which is a slightly strange thing to be doing a fortnight after their industry was in the papers for the wrong reasons. The argument hasn’t changed. It’s just been given a very expensive illustration.

The advice I gave those five rooms in 2018 was that in a world of routine and technology, their wisdom would become the most important thing they had. I still think that’s right. What I’d add now, eight years and one Operation Claw later, is that being the trusted human in the room is not a position you get to occupy passively. It’s a claim, and claims attract counterfeit, and the only durable defence against counterfeit is being demonstrably, specifically, unrepeatably yourself with the person in front of you.

That is a lot harder than putting trust on the wall. It’s also the only part of this that a machine, a network, or a very good forger cannot supply.

I’ve written a longer piece of research on where this is heading for the broking profession specifically, and I’m bringing it into that room on the day. If you’d like a copy afterwards, say so and I’ll put you in the queue.

Choose Forward.

---

Morris Misel is a foresight strategist and keynote speaker. He works with leaders, boards and associations on what technological, social and economic shifts mean for people, organisations and decision making.