Bendigo Bank - 1,598 Accounts, One Password, and the $8 Million Bill That Arrived Three Years Later
21 minAudio
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! On 11 August 2026, Bendigo and Adelaide Bank accepted a proposed $8 million penalty over a March 2023 cyber attack on Service One Alliance Bank. The court documents contain a sentence worth reading twice: at the time the attack began, 1,598 customer accounts were protected by the password 123456. The timeline. 2020: penetration testing identifies security weaknesses in the Alliance Bank environment. Not remediated. October 2022: a threat actor attempts brute-force login attacks and fails; the board is told the bank is at a critical point of needing further investment in resourcing and capability. No substantive review is undertaken. March 2023: an unidentified attacker brute-forces approximately 257 customer accounts. 10 August 2026: APRA files an Originating Application in the Federal Court. 11 August 2026: APRA announces publicly and Bendigo lodges an ASX announcement accepting the proposed penalty. The numbers - all official, from APRA, the Federal Court filing and Bendigo's ASX announcement, none of them threat-actor claims: 257 accounts accessed; 286 unauthorised transactions; 87 customers affected by a transaction; approximately $490,000 misappropriated; about $140,000 unrecoverable; all customers fully reimbursed; 1,598 accounts using 123456; an $8m proposed penalty; and about $2.6m in additional legal costs. The three named control weaknesses: password settings permitting very weak passwords; multiple accounts protected by identical passwords; and system design that enabled the attacker to identify valid customer IDs - account enumeration. The regulatory angle. This is not an OAIC or Notifiable Data Breaches matter, because no bulk personal information was taken. The regulator is APRA and the instrument is the Banking Executive Accountability Regime. All four admitted failures are governance failures: inadequate authentication controls, no systematic testing programme, accountable persons' responsibilities not covering the Alliance Bank IT system, and inadequate information security governance. Attribution: none. APRA and the court documents describe an unidentified threat actor. No group claimed it, no data was published, no ransomware was involved. Five lessons. One: a penetration test you don't remediate is worse than no penetration test - it is a dated written record that you were told. Two: treat a failed attack as a real incident; October 2022 was a rehearsal nobody reviewed. Three: password policy isn't solved because you assume it is - go and measure it, and make MFA default rather than optional. Four: account enumeration is a vulnerability, not a UX detail - identical responses and identical timing, whether or not the account exists. Five: outsourced, white-labelled or subsidiary IT still carries your accountability, so put a person's name, not a team's, against every system your customers touch. The penalty is roughly 16 times the amount stolen. It isn't for the loss; it's for the three years between the finding and the fix. Visit www.kinsoft.com.au to talk through your security and IT needs. Sources: APRA media release, Bendigo and Adelaide Bank admits breaching its BEAR obligations, 11 August 2026; APRA Originating Application, Federal Court of Australia, stamped and redacted, 10 August 2026; Bendigo and Adelaide Bank ASX announcement, 11 August 2026; Information Age (Australian Computer Society), 12 August 2026; Cyber Daily, 11 August 2026; FST Media; Lawyerly; Mortgage Professional Australia.
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
Picture this, if you will. It is March 2023. Okay. And a cyber attack on a pretty major Australian bank is just kicking off. But at that exact moment, 1 ,598 customer accounts at this institution are secured by a single identical password. Oh, wow. Yeah. And that password was literally 123456. I mean, that is just is painful even here. It really is. So welcome to today's deep dive. We are unpacking a really fascinating August 2026
episode of Tech Talks with Kinsoft. Right, which goes into all the details of this breach. Exactly. And our mission today for this deep dive isn't just to, you know, gawk at terrible security hygiene, though there is plenty of that. Oh, yeah, there definitely is. But we are really analyzing this massive regulatory paradigm. shift because this relatively small theft of about $490 ,000 resulted in a crushing $8 million penalty. Yeah, which is just a staggering scenario to encounter in modern finance. I mean, when you hear a statistic like nearly 1 ,600 accounts using 123 -456, and this is in the year 2023,
you know? Yeah, it's not 1998. Right. It completely shatters our baseline assumptions about institutional competence. Oh, absolutely. We naturally assume the foundational walls of a heavily regulated entity are built of reinforced concrete. But in this specific corner of the bank, there wasn't even a screen door. OK, let's unpack this, because to really grasp how regulators arrived at an $8 million penalty for a fraction of that and actual losses, we have to look at the mechanics of the breach itself. We do, yeah. So the Australian
Prudential Regulation Authority, or APRA, they identified three specific vulnerabilities. And I'm telling you, they defy every assumption you probably have about modern banking. You look at it. So first, obviously, the complete lack of a restriction. Right. There were just no guardrails at all. None. And second, they allowed multiple accounts to share identical passwords. Which, you know, fundamentally changes the math for an attacker. How so? Like for the listener who
isn't a pen tester. Well. You no longer have to guess the correct password for a specific user. You just take the most statistically common password in the world and you simply spray it across the entire user base. Right. You just try the same key on every single lock. Exactly. You guess once and you apply it everywhere. Right. But then there's the third vulnerability, which is the one that actively assisted the attackers, honestly. Oh, yeah. Account enumeration. Yes, account enumeration. For you listening, I want you to think of it this way. It's like a burglar. Walking down a long, dark hotel hallway, just
jiggling door handles. And every single time they touch a handle, the door loudly announces someone lives here or this room is empty. That's a great analogy, actually. Thanks. I mean, the login process for this banking platform was essentially doing exactly that for the attacker. It was confirming exactly which user identifiers were real and which ones were completely fake. Yeah. And the source material has this brilliant phrase about this. It says these three flaws combined turned a search problem into a shopping list. A shopping
list. I mean, that is just it's chilling, honestly. What's fascinating here is. These are not sophisticated zero -day exploits. Right. No one is writing custom malware here. Exactly. They are depressingly ordinary everyday user experience decisions. Wait, UX decisions? Yeah. Think about it from the perspective of a UX designer. Like, when a user forgets their login, the designer wants to be helpful. Sure. You don't want the user to get frustrated. So they program the system to explicitly say, we don't have an account under
that email address. Oh, right. Because it feels like good customer service. Exactly. But from a security standpoint, you are actively doing the reconnaissance for the threat actor. Wow. Yeah, you're literally handing them the map. Right. Because if an attacker doesn't have enumeration, they are just blindly throwing both usernames and passwords into an infinite void, just hoping they eventually line up. But with enumeration, the void just disappears. It does. They now have a confirmed, validated list of actual customer IDs. It's just wild. I mean, the math literally
stops being hard. You have the exact usernames and you know almost 1 ,600 of them are using the six most predictable characters on the keyboard. Yeah, it's barely even hacking at that point. No, it's just logging in. But if the attack was that elementary, someone inside had to see it coming, right? You would think so. And looking at the timeline from the source material, they absolutely did. Yeah, the timeline is precisely where this narrative shifts. It goes from a technical failure to a catastrophic governance failure. Okay, so let's look at this three -year window
outlined in the source. In 2020, three full years before the breach, a penetration test successfully identified these exact security weaknesses in the environment. Three years. Three years, but they weren't remediated. Then, jump forward to October 2022, an actual threat actor attempts a brute force login attack on the platform, and they fail. Right, the system actually held them off that time. Yeah. And following that failed attack, the banks board was explicitly told that they were at a, quote, critical point of needing
further investment in resourcing and capability. But no substantive review was undertaken. Nothing. Five months go by. And then in March 2023, an attacker comes back and this time they walk right through the front door. You know, that October 2022 event, it really cannot be overstated. Because they knew someone was knocking. Exactly. In the security world, a failed attack is essentially free intelligence. Free intelligence. Yeah, it is a massive gift because it tells you exactly what somebody wants. It shows you the methods
they're using to get it. And most importantly. It tells you they've chosen you. Exactly. It tells you that they have specifically chosen you as a target. The alarm bells were absolutely ringing. Well, okay, let me play devil's advocate for a second here. Sure. Because I want to look at this from the perspective of the IT teams working in the trenches. Realistically, enterprise security operations centers, they get thousands, maybe tens of thousands of automated alerts every single day. Oh, yeah. The noise is deafening. Right. The Internet is constantly scanning and probing literally everything with an IP address.
So if an automated script fails to log in, isn't that just like a regular Tuesday? That's a fair point. Like how is a massive bank supposed to overhaul its entire infrastructure every time a brute force attempt fails? And that is a completely valid defense if we're talking about an isolated alert that just got buried in a log file at the analyst level. Because it's easy to get lost in the noise. It is incredibly easy to get lost in the noise. But that defense totally evaporates when you look at the source material for this
specific case. Right. Because of who was told. Exactly. The highest levels of leadership were briefed. The board was told in the direct aftermath of that failed attack. that the system was at a critical point. Wow. The warning successfully traveled all the way up the chain of command, bypassing the noise of the SOC. And yet, nothing structural changed over the next five months. Just crickets. Right. And knowledge is only valuable when it is applied. Having a 2020 penetration test report sitting in a digital folder somewhere,
you know, that isn't security. No, not at all. It is simply a dated, written record of your negligence. Wow. A dated written record of negligence? I mean, you'd think after a failed brute force attack that actually reached the board level, someone would just, I don't know, pull the plug or force a mandatory password reset for the entire user base. You would certainly hope so. But no one did. And reading through the source material, I noticed something strange about the corporate structure that might actually explain why nobody acted. Ah. Yes. You're referring to the governance
gap. Yes. I was looking at these notes and getting a bit confused, honestly. So the parent company here is Bendigo and Adelaide Bank, which is Australia's sixth largest bank. You're right. They're the ones who got fined. But the customers who actually got hit belong to ServiceOne Alliance Bank. So how does a massive, heavily regulated banking system operate a platform where the legal entity and the operational entity are so detached? Well, it is a classic white label arrangement, and it creates a massive structural blind spot. Okay,
break that down for us. So Bendigo and Adelaide Bank held the overarching banking license. Therefore, Bendigo carried all the regulatory accountability. Okay, that makes sense. But they provided a digital banking platform to a network called Alliance Bank, which was made up of several authorized representatives, including ServiceOne. Got it. So that basically rented out the tech. Essentially, yes. Yeah. But the platform those Alliance customers used fell into this bizarre, undocumented middle ground. Like a no man's land. Exactly. It was
a system serving Bendigo's regulatory obligations, but nobody within Bendigo's IT or risk departments had been explicitly told this specific platform is your problem. Oh, geez. It was an orphan system. See, here's where it gets really interesting. Because of this orphan system. When Bendigo ultimately faced the regulator, they admitted to four specific failures. And if you read them, not a single one of those failures is, we got hacked. No, they aren't. They admitted they failed to maintain authentication controls. They failed to test those controls. They failed to ensure accountable
persons covered the IT system. And they failed to maintain risk management governance. Right, all governance. I mean, how does a modern enterprise... Just not have a specific human being's name attached to a live banking environment. Well, it happens far more often than you would think, unfortunately. Especially in the wake of acquisitions, white label partnerships, or complex managed service contracts. It just gets lost in the shuffle. It does. And if we connect this to the bigger
picture. This structural void is exactly why the regulatory response was so unique. The regulator in this case wasn't the usual privacy or data protection commission. Right, because there was no mass exfiltration of personal data, right? Like no database of passports or driver's licenses was just dumped on the dark web for sale. Precisely. The regulator was APRA, and the instrument they used was BEAR. BEAR, which stands for? The Banking Executive Accountability Regime. Okay, so it's focused on the people at the top. Exactly. BEAR is a framework explicitly designed to force personal
accountability onto senior executives and directors. So they can't just hide behind the corporate logo. Right. The punishment here was strictly and entirely about the failure of governance. It was about the fact that leadership could not point to a single person responsible for the platform's security posture. Which brings us to the math. Because that absolute void in accountability is what triggered a financial penalty that completely destroys traditional enterprise risk modeling. It really does throw the old models out the window.
Let me give you the numbers punchily for anyone listening. The attacker gained access to 257 accounts. Right. They made 286 unauthorized transactions. In total, about $490 ,000 was stolen. Which in banking terms is relatively small. Exactly. Now the bank fully reimbursed every single affected customer. The bank itself, as a multibillion -dollar entity, was never financially threatened. Not at all. But the penalty handed down by the federal court, $8 million. Plus another $2 .6 million in legal costs. Right. Over $10 million
total for a $490K loss. And that arithmetic right there is the most critical takeaway for anyone listening who has to justify a security budget or assess corporate risk. Well, I have to push back on this, though. Okay. Let's hear it. I'm looking at this from the perspective of a CFO or a risk manager. If I build a quantitative risk model, I look at this platform and say, worst case statistical scenario. We lose half a million bucks to fraud. Statistically, yes. Right. So when the IT director asks for a $2
million budget to overhaul the white label platform, I decline the spend. It makes pure financial sense. Under the old way of thinking, sure. So getting hit with an $8 million penalty for a $490K loss, I mean, doesn't that sound like massive regulatory overreach? If I'm an executive, I'm appealing that as absurdly disproportionate. It's like getting a speeding ticket that costs more than the car you were driving. It is. Your analogy is spot on, but your risk model is exactly what the regulators are trying to break here. Okay. Explain that. The $8 million penalty was
not a punishment for losing $490 ,000. It was a punishment for the three years of dead air between the penetration test finding the flaw in 2020 and the fix that didn't happen until after the breach in 2023. Ah, so it's a penalty for the silence. Exactly. Regulators under the Bayer framework are drawing a very hard, deliberate line. If an unknown, highly sophisticated zero -day vulnerability is exploited by a state -sponsored actor, that is considered bad luck. Right, because nobody could have seen it coming. Yes, you mitigate
it, you report it, and you move on. But the cost of a known, documented, and internally reported vulnerability that you simply chose not to fix. That's different stories. Completely different. That is no longer bad luck. That is an $8 million corporate decision. An $8 million corporate decision? Wow. That phrasing completely changes how you have to view a backlog of IT tickets. It really should. So how do you, the listener, actually avoid making an $8 million decision in your own organization? The source material lays out five incredibly clear lessons. Let's really dig into
these because they apply whether you are running a startup, managing an IT department, or just trying to be a more informed professional. Sounds good. Let's go through them. Lesson number one. An unremediated penetration test is worse than no penetration test at all. Yeah. And this is a harsh reality of legal discovery. Yeah. As we discussed, an ignored pentest stops being a proactive security tool and immediately becomes legal evidence against you. Right. You're basically doing the prosecutor's job for them. Exactly. When a breach happens, the very first thing regulators
and opposing counsel will ask for is your history of external audits. If they find a PDF from three years ago outlining the exact vulnerability that was just exploited, you have lost all deniability. There is no we didn't know. None. If you commission a test, you must have a named accountable owner for every single finding, a strict due date for remediation, and an aging report that gets escalated to the board if those dates are missed. If it just sits in a drawer, you are manufacturing your own liability. That makes perfect sense.
Okay, moving to lesson number two. Treat failed attacks as real incidents. We touched on this earlier, but what does that practically mean for a security team? Well, it requires a fundamental shift in mindset. Many organizations view a blocked attack as a success. Right, like, hey, the firewall held, the login failed, we're good. But a failed brute force attempt, especially a sustained one, is a rehearsal. It requires a formal incident review with a concrete outcome. Like you have to ask the hard questions. Yes. What were they
targeting? Where did the traffic originate? Did they pivot to another subsystem when the front door held? If your infrastructure can't even provide the visibility to answer those questions after a massive spike in failed logins, you don't have a security program. You just have a very busy firewall. Which brings us to lesson number three. Actually measure your password policy. Don't just assume it works because HR sent out a memo. Oh, this is a big one. Everyone claims to have a robust password policy. But you have to measure its reality against user behavior.
Are you actively checking your Active Directory against known breached password lists? And more importantly, let's talk about multi -factor authentication, or MFA. Right. The source material explicitly mentions the failure of MFA here. Yes. And the lesson is brutal. It is. If MFA is available but optional, you must assume, statistically, that it is completely off. Optional equals off. Exactly. User friction is the enemy of adoption. If you give a user the choice to skip entering a six -digit code on their phone, the vast majority
will absolutely skip it. Yeah, I've skipped it myself when I'm in a hurry, honestly. We all have. But you cannot tell a regulator, well, we offered them the security tool. Security isn't an opt -in newsletter. If it protects sensitive data, it must be mandatory. Available but optional equals off? That is a fantastic rule of thumb for any product manager to keep in mind. Okay, lesson number four. Fix account enumeration. Right. We talked about how the system told the attacker which accounts were real. But how do you actually fix that without ruining the user
experience? It requires a bit of technical discipline. The lull in, password reset, and account registration flows must return identical responses, regardless of whether the email address exists in your database or not. So no more email not found pop -ups. Right. Instead, the system should say, if an account exists for this email, a password reset link has been sent. But... You know, it goes deeper than just the text on the screen. Wait, how so? Timing attacks. Timing attacks, like literally measuring time. Yes. If your database takes 50 milliseconds to realize a user doesn't
exist, but it takes 500 milliseconds to verify a password for a user that does exist, an automated script will measure that time difference and enumerate your database anyway. Oh, wow. I never even considered that the processing time itself could give away the existence of an account. Most people don't. Your engineering team has to ensure uniform response times across all authentication flows. It is a remarkably cheap fix at the code level, but almost no one tests for it. And fixing it completely denies attackers that validated
chopping list we talked about earlier. That is fascinating. Okay, finally, lesson number five. Outsourced IT still carries your accountability. And this is the governance trap that caught Bendigo. You can outsource the technical labor. But you cannot outsource your regulatory or reputational responsibility. You can't just point fingers at the vendor. Exactly. If a managed service provider runs your infrastructure, you own that risk. If you acquired a smaller business three years ago that is still quietly running its old legacy tech stack in the background, you own
that risk. If you use a third -party software as a service or SaaS platform that is branded with your company's logo. You own that risk. You own that risk. So you essentially need a map of everything. You need a map and you need names. Yes. You must be able to write down every single digital system your customers interact with and put one specific executive's name next to it. Not just the IT team. No, not a whole department. One person. Because, as the source material proves, if you don't proactively pick someone to be accountable, the regulator will
gladly step in and pick someone for you. What does this all mean for you listening? I want you to genuinely visualize your own workplace right now. Think about the software you log into every day. Think about the vendors your company partners with. Do you have a SaaS platform handling client data or maybe an old acquired business unit operating under your brand that no specific human being actually owns? Because if you look closely and you realize there is an orphan system sitting out there in the digital ether, Well, right now, that is a massive, multimillion -dollar
blind spot. Yeah, and just to summarize what we've unpacked from this source material today, we have officially crossed the threshold into a new regulatory era. We really have. Regulators are no longer just waiting to punish companies for losing massive databases of personal information. They are actively and aggressively punishing companies for failing to govern their own known risks. Even if the actual data loss is small. Right. The absence of internal oversight? The lack of a named owner? The ignored audit report? These are now treated as severe compliance failures
in their own right, regardless of the direct financial impact of the breach itself. It is a whole new world of accountability. Thank you so much for joining us on this deep dive. We've covered a tremendous amount of ground, from burglars in hotel hallways to the intricacies of the federal court and timing attacks. But I want to leave you with one final lingering thought to mull over today. Let's hear it. The source material makes it abundantly clear that a three -year -old security... finding with no designated owner isn't an oversight. It's a deliberate choice. So thinking about your own organization, your
own department, or the projects you manage right now, what decisions are currently sitting quietly in your backlog just waiting for the right attacker to find them? Because you definitely don't want the answer to be one, two, three, four, five, six.
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Japanese home goods company Iris Ohyama (online store operator Iris Plaza) found a suspicious file in an external...
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Australian online retailer Who Gives A Crap told customers on 7 November 2026 that their information was accessed...
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Last week in tech, 7 to 14 November 2026. Microsoft's November Patch Tuesday (10 November) fixes well over a...
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Chainalysis disclosed on 30 October 2026 that an external actor using "offensive agentic techniques" accessed...
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! NSW Health reported unauthorised access to its Oral Health Fee For Service Scheme website on 29 October 2026;...