McGraw Hill – 13.5M Exposed via Salesforce Misconfig
18 minAudio
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Edtech giant McGraw Hill confirmed a breach (dated around 10 April 2026, surfacing mid-April) after ShinyHunters' extortion threat; about 13.5 million accounts were exposed - emails, names, phone numbers and physical addresses - via a Salesforce misconfiguration. When negotiations failed, the group published over 100 GB of data. McGraw Hill said the exposed data was limited and did not include Social Security numbers, financial information or student data from its learning platforms. We discuss misconfiguration risk and the limits of non-sensitive data. Need a configuration and exposure review of your SaaS estate? Visit www.kinsoft.com.au to talk through your security and IT needs. Sources: BleepingComputer; The Register.
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.
Welcome to Tech Talks with Kinsoft. Glad to be here. So today's mission is to unpack a really massive cybersecurity failure. And I mean, we want to extract the most vital lessons. about modern software risks from it yeah it's it's a huge topic it really is we're looking at how organizations manage those complex often completely invisible connections between the software they use every single day right and what happens when those connections are just neglected exactly
yeah so i want you to imagine for just a second that you're in charge of building an impenetrable digital fortress. Right. You spare absolutely no expense. You got the moats, the armed guards, all of it. Yeah, the state -of -the -art laser grids covering the front door. You are entirely secure. But then, completely by accident, you leave a hidden ground floor side window, just wide open. Oh, man. And not just open, but, like, propped open for literally anyone walking past to see. Right. And this scenario, it isn't some hypothetical nightmare. This is exactly what happened to the K -12 and professional education
giant McGraw -Hill in April 2026. Yeah. And it highlights a really painful reality in modern IT infrastructure. You know, you can have the strongest front door in the world. I mean, fortified by multifactor authentication, cutting edge firewalls, the works. Right. But if that side window is left open, threat actors, they don't need to pick a lock. They don't need to deploy some sophisticated malware. They just walk in. They just climb right in. And the resulting fallout is usually staggering. Staggering is really the only word for it. I
mean, we are talking about. A data breach exposing 13 .5 million accounts. Yeah. And to really grasp the scale of this failure, we first need to look at the timeline, right? The timeline of the attack and the threat actors who actually pulled it off. Definitely. So the breach surfaced in mid -April 2026. That's when this extortion group known as Shiny Hunters. Right. Well -known group. Yeah. They paraded McGraw Hill onto their dark web leak site. Yeah. And they issued a ransom demand. And it had a strict April 14 deadline
attached to it. Right. And that use of a public countdown timer, that's a very specific psychological tactic. Oh, really? How so? Well, it's designed to bypass the normal, rational incident response procedures of a corporation. It just creates sheer panic in the boardroom. So they just want to cause chaos. Exactly. They want executives making rushed, fear -based decisions under extreme public pressure. Right. Just hoping they pay up. Yeah. Hoping someone just authorizes a payout to make the whole PR nightmare disappear. But McGraw -Hill, apparently they didn't cave to
that pressure or maybe the negotiations just completely broke down behind the scenes. Right. Because when that April 14 deadline passed without a payment, shiny hunters actually followed through on their threat. They dumped over 100 gigabytes of data online. Which is a massive amount of information. It's huge. But looking at the source reports, there's this fascinating detail. Shiny Hunters didn't just list McGraw -Hill on this leak site. They also listed Rockstar Games. And they claim to be holding over 40 million Salesforce records containing PII. Personally identifiable
information. Right. Highly personal stuff. Which kind of makes me wonder about the targeting here. Does this mean McGraw -Hill wasn't specifically targeted? That's the million dollar question. Because McGraw -Hill is a massive institution. They have millions of users across schools and corporate training. Why wouldn't they be a high value specific? It just feels like a goldmine, not just some random catch in a dragnet. And if so, how does that change how we should view corporate security? Right. Well, it is a goldmine. But the methodology of how they acquired it,
that points directly to a dragnet approach. Explain that. If you look at Shiny Hunter's history, especially a really similar campaign they ran back in 2025, you see their operational model. Right. They aren't spending months trying to crack the... intricate, bespoke security systems of individual companies. That takes too much effort. Exactly. Trying to break down the front door of a company like McGraw -Hill requires massive resources, zero -day exploits, highly specialized skills. The return on investment
for the hackers just isn't there when there's an easier path. So if they aren't spending months cracking the core system, they have to be finding a skeleton key somewhere else. Bingo. How does a dragnet actually work in this context then? So they automate the discovery phase. Instead of targeting a specific company, they target a specific vulnerability or a misconfiguration in connected cloud services. Okay. They write scripts that just constantly scan the entire internet. They're pinging millions of servers
and API endpoints looking for one specific flaw. Wow. Just sweeping the web. Yeah. And in this case, they were looking for misconfigured Salesforce environments or exposed integrations. So they really do just cast a massive net. Exactly. If you happen to have that misconfiguration, your data gets scooped up automatically. They don't even care who you are. No, they don't care if you're an education publisher like McGraw -Hill or a video game developer like Rockstar Games. To them... Data is just leverage. Man, that really shifts the entire paradigm of how we think about
risk. You don't have to be a high -value target to get hit. Not at all. You just have to be making a really common administrative mistake, which honestly brings us to the actual technical mechanism of this brief. Right, the side door. Yeah. According to the reports, McGraw -Hill claimed the leak stemmed from a, quote, limited Salesforce -hosted web page. Notice the phrasing there. Right. They were incredibly quick to insist that there was absolutely no unauthorized access to their core Salesforce accounts, their customer databases,
courseware, or internal systems. Right. They even went so far as to state that this appears to be a broader issue involving a misconfiguration within Salesforce's environment that has impacted multiple organizations. Yeah, that right there is a total masterclass in corporate crisis communication. It really is. Notice that careful use of the phrase within Salesforce's environment. Yeah. They are actively attempting to shift the perception of blame right onto their vendor. It's essentially like hiring the best security firm on earth to
build a bank vault, right? Right. You have the titanium doors, the time locks, the laser sensors. But then one of your own employees. accidentally tapes the architectural blueprint of the vault, complete with the override codes, to a public billboard out on the highway. Exactly. The bad guys didn't break your vault. The vault worked perfectly. They just bypassed it entirely because you handed them the schematic. That's a great way to put it. And what really raises major questions about this figure pointing is that when journalists reached out to Salesforce for a comment on this
supposed broader issue. Yeah, what did Salesforce say? Salesforce gave no response. They totally ghosted the media. Right. And that silence from Salesforce speaks volumes about the shared responsibility model. Which is what exactly? Well, it's arguably the most misunderstood aspect of cloud computing today. OK. When you use a massive platform like Salesforce or Amazon Web Services, Microsoft Azure. Yeah. The vendor is responsible for the security of the cloud. Okay, so the physical stuff. Yes, they secure the physical servers,
the data centers, the network infrastructure, and the underlying code of the application itself. But the customer, McGraw -Hill in this scenario, is responsible for security in the cloud. Meaning how they actually use the tool. Exactly. The customer manages their own access controls, how they configure public -facing portals, who they give administrative rights to, and crucially, what third -party applications they integrate into the system. Oh, the integrations. Yeah. Most compromises we see on these platforms do
not stem from a flaw in Salesforce's code. They come from the users. Yes. They stem from user end issues, stolen credentials, poor access management, or abused OF apps. Okay, wait. Before we get into the integrations themselves, what exactly is an OF app? Is that like the login with Google or login with Salesforce button I see everywhere on the web? That is exactly what it is. Oh, really? Yeah. OF is an open standard for access delegation. It's what allows you to grant a website or application access to your information. on another website
without giving them my actual password exactly so when a marketing tool says connect to salesforce to import your contacts it uses of it asks salesforce for an access token if a mcgraw -hill administrator clicks approve that marketing tool now has a token that acts as a persistent invisible bridge into the salesforce database wow and if i'm understanding this correctly The problem isn't the bridge itself. The problem is how wide you make the bridge and who you let drive across it, right? Precisely.
We're talking about over -permissioned integrations. That is the core issue here. Think of an integration token like a valet key for your car. Okay. You give the valet key to the parking attendant so they can park the vehicle. It should only start the ignition and open the driver's door. Right. Nothing else. But if you accidentally give the valet a master key, that also opens the glove box, unlocks the trunk, and contains the code to your home security system. Oh, man. Yeah, you have over -permissioned that key. If that valet key is stolen, the thief has access to
everything. But why do companies do this? If IT departments know that over -permissioning is dangerous, why do these master keys get handed out to random marketing tools or connected web pages? It usually just comes down to the friction between business convenience and cybersecurity. The classic battle. Always. Marketing teams, sales teams, customer support, they need to move quickly. They want to launch a new portal or connect a new analytics dashboard today. Right. But the IT security team is often backlogged.
So instead of taking the time to painstakingly configure an integration so it only has read -only access to three specific fields. Someone just clicks up. Prove all. Basically. An administrator might just check the box that grants global read and write access to the entire database. It makes the integration work instantly. Which makes the business side happy. Exactly. But it creates a massive silent vulnerability. Because to the system, when that integration asks for data.
It looks entirely legitimate. It looks completely normal. The integration has a valid token. It was granted permission by an administrator. So it just lets them in. Yeah. So when the attacker compromises that third -party tool or finds an exposed portal utilizing that token, the system says, sure, here are the 100 gigabytes of data you requested. That is terrifying. It turns the ability to connect software seamlessly into a devastating weapon. The attacker pulls the data without ever triggering an intrusion alarm on the core systems. The systems McGraw -Hill was
so eager to defend. Right. The data just leaks out the side window while the armed guards at the front door see absolutely nothing. Exactly. Which really brings us to the human cost of this failure. And honestly, the dizzying corporate spin regarding what was actually stolen. Oh, the spin was intense. Yeah, because... McGraw -Hill immediately deployed a communication strategy focused entirely on what wasn't taken. Right. They announced that no social security numbers were compromised, no financial account information,
no student data from their learning platforms. It's the standard playbook for incident response, really. Just calm everyone down. Yeah. The immediate objective is to stop the bleeding of public trust. And the easiest way to do that is to assure people their bank accounts are safe. Which, I mean, I'm certainly glad my credit card isn't floating around the dark web. However, the security tracking site have I been pwned independently confirmed the scope of the 13 .5 million exposed records. And it was a lot. It was. That data included
full names, personal phone numbers, email addresses, and actual physical home addresses. Yeah. And yet. McGraw -Hill categorized this as a limited exposure of, quote, non -sensitive data. Right. I have to say, if a criminal syndicate has my full name, the email I use for everything, my personal cell phone number, and the exact physical address where I sleep at night. You don't feel secure. I don't feel like the exposure is limited. I feel entirely compromised. Absolutely. At what point do we as an industry stop letting companies
use the phrase nonsensitive data as a get out of jail free card to dismiss massive breaches? Well, you are completely justified in feeling compromised. The phrase nonsensitive is a dangerous myth. Thank you. It really highlights a severe disconnect between corporate liability and individual risk. Oh, so? Well, legally and historically, corporations fixate on social security numbers, health care records and credit card numbers. Right. Because those specific data points carry direct regulatory fines and immediate financial liability. If a credit card is stolen, fraud
occurs. Banks initiate chargebacks. So the company actually faces measurable immediate damages. Exactly. The company gets punished. So they care about. that specific data but with names and addresses they don't right but a hundred gigabyte database of verified names active emails personal phone numbers and physical addresses yeah That is the ultimate starter kit for modern cybercrime. Wow. It's the raw material needed to launch devastating secondary attacks. It might not cost McGraw -Hill money today, but it shifts an enormous burden
of defense directly onto the individuals. The people whose data was actually exposed. Yes. So walk us through how that secondary attack actually plays out. If a threat actor buys this McGraw -Hill data set off the dark web, what do they actually do with my physical address and my phone number? They use it to craft highly - targeted, socially engineered attacks that just completely bypass your natural skepticism. Okay, give me an example. Let's say you receive a text message on your phone. It addresses you by your first and last name. Cool. It states there's an issue with a package delivery or a
municipal utility notice. And it actually includes your physical home address right in the text. Oh, wow. Yeah, and it asks you to click a link to verify delivery instructions. See, my brain immediately drops its guard there. Because if I get a generic text saying your package is delayed, I usually just ignore it. Right. Everyone does. But if the text says, we cannot deliver. to a 123 main street and that is my actual house do you pay attention the context makes it feel totally authentic exactly it leverages familiarity to
manufacture trust so you click the link which takes you to a spoofed login page for a delivery service or your email provider and i just hand over my password you do or consider how this data is used against your employer attackers use this data set to profile you oh right they call your company's i .t help desk pretending to be you. They say they need to reset your password. And the help desk asks security questions. Right. And when the agent asks for verification, the attacker has your home address, your personal
phone number, your personal email address, all ready to go. Because they bought it from the McGraw -Hill leak. Yes. They use that data to authenticate themselves as you, gaining access to your corporate network. Man, the danger really just compounds. It does. I mean, it's technically true that McGraw -Hill didn't lose your social security number or your password. Right. But they handed attackers the exact puzzle pieces needed to trick you into handing those things over yourself. And that is the reality of the threat landscape today. When an organization
tells you a breach only involved non -sensitive contact information, you just have to assume that your details are now being actively used to profile you. For sophisticated scams. Yes. You have to become hypervigilant about every unsolicited text, email or phone call that uses your personal information to try and establish credibility. Which is an incredibly heavy burden to carry, especially when you trusted the institution to begin with. Absolutely. And honestly, looking at the entire scope of this breach, there is a profound irony here that is just difficult
to ignore. Oh, for sure. We're talking on McGraw Hill. I mean, this is an institution built entirely on. digital learning platforms they provide assessments educational software textbooks every spanning from k -12 classrooms all the way through to professional corporate training environments they are fundamentally in the business of teaching right and yet they just received a very public very painful and very expensive lesson in digital security yeah It serves as the ultimate cautionary tale for any organization operating today. Really?
Yeah. It proves that no matter what your core product is, whether you publish textbooks, develop video games, or sell shoes, if you operate in the modern world, you are a technology company first. That's a great point. And if you do not understand the complex web of integrations, APIs, and permissions that make up your digital infrastructure, you are deeply vulnerable. And as we've seen today, that vulnerability isn't always some dramatic hack with a team of cyber criminals deciphering firewalls in a dark room. Usually it isn't. Sometimes it's just a misconfigured setting on a web page.
Or an over -permissioned integration token that some administrator approved to save time and then completely forgot about. Which happens every single day. Right. So what does this all mean for you listening? I want to leave you with a thought to mull over, extending beyond just the headlines of this specific breach. Okay. Think about all the third -party apps connected to your own work email or your company software right now. The calendar integrations, the project management tools, the marketing plugins, the simple web portals. There are dozens of them.
At least. If a tech -reliant titan like McGraw -Hill can accidentally leave a digital window wide open for anyone to find, how many invisible over -permission connections are quietly sitting in your own organization's tech stack, just waiting for an automated scanner to discover them? It's a scary thought. It really is. To discuss your security and IT needs, visit www .kinsoft .com .au.
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;...