All Articles
Johnny Collins
In the News

No Malware. No Breach. Just a Bank Employee and $3.4 Million in Fraud.

A former TD Bank employee pleaded guilty on May 28, 2026 to helping facilitate more than $3.4 million in fraud, according to a recent U.S. Department of Justice announcement. The case raises a challenge that organizations across every industry should be asking themselves: how does fraud committed by trusted insiders using legitimate systems go undetected long enough to reach millions in losses?

‍

When Legitimate Systems Become Fraud Tools

‍

One of the most overlooked realities of insider fraud is that it rarely requires hacking.

‍

Employees simply misuse applications, portals, and systems they are already authorized to access. And much of that misuse never leaves a trace in the places security teams are trained to look.

‍

A bank employee searches for high-value accounts using approved banking software. A customer service rep pulls up records outside their assigned responsibilities. A healthcare worker views patient files unrelated to their job. An insurance employee repeatedly opens sensitive claims data with no clear business reason.

‍

From a technical perspective, these actions look completely legitimate. The user is authenticated. The application is approved. The access is permitted.

‍

But what's actually happening on screen tells a different story.

‍

The employee isn't just accessing the system. They're navigating to specific account types, running searches designed to surface high-value targets, reading data they have no business reason to view, and doing it repeatedly across multiple sessions. None of that appears in an access log. None of it triggers a firewall alert. It's just a person using an application the way the application was designed to be used.

‍

That's what makes it so hard to catch.

‍

‍

The Proprietary System Problem

‍

The challenge deepens significantly when fraud occurs inside proprietary or internally developed applications.

‍

Many organizations rely on custom-built portals, line-of-business systems, and internally developed tools to handle their most sensitive operations. These systems often contain the highest-value data in the organization and have no integration with third-party security or monitoring tools.

‍

Standard DLP solutions can't see inside them. Fraud detection platforms have no context for what's happening within them. SIEM tools may log that a user authenticated, but have no visibility into what that user actually did once they were in.

‍

When an employee misuses an on-screen component in a proprietary system, such as running an unauthorized search, navigating to restricted record types, or extracting information by reading and memorizing it rather than downloading it, there may be no signal at all reaching the security stack.

‍

The activity is invisible. And invisible activity is exactly where fraud takes root.

‍

Why Traditional Fraud Controls Miss This

‍

Most fraud detection platforms focus on completed transactions. Most DLP solutions focus on data movement through email, web uploads, cloud services, or removable media. Security monitoring tools focus on logs, authentication events, and technical indicators.

‍

Each serves an important purpose. But many insider fraud schemes begin long before any suspicious transaction occurs and without any data ever leaving the organization in a detectable way.

‍

An employee gathering account information for future misuse may never trip a transaction monitoring alert. A user repeatedly reviewing sensitive customer records may never generate a DLP event. A fraudster operating entirely within an authorized application, reading information off a screen and passing it along verbally or through personal devices, may never look suspicious from a traditional security standpoint.

‍

By the time a fraudulent transaction occurs, the planning and information gathering may have been running for weeks or months.

‍

What Actually Happened at TD Bank

‍

According to prosecutors, former employee Cheungkin Lam (also known as Kelvin Lam) admitted to accepting bribes and abusing his bank position to identify customer accounts with large balances. He provided confidential customer information to external fraudsters, enabling schemes that resulted in more than $3.4 million in losses. He also participated in a separate effort involving bribing an employee at another financial institution to falsify records for additional fraud.

‍

The guilty plea is the latest development in a case that shows how some of the most damaging fraud incidents don't start with external attackers. They start with employees who already have access to everything they need, using systems they're fully authorized to use, doing things that look completely normal until they aren't.

‍

Detecting Fraud Before the Transaction

‍

The behaviors that often precede fraud aren't always found in transaction logs or network activity. They happen inside applications, through user interactions that look routine when viewed in isolation. And in many cases, they happen entirely on screen, in ways that only become meaningful when you can actually see what the user was doing.

‍

InnerActiv was built around this reality.

‍

Rather than focusing exclusively on completed transactions or data movement, InnerActiv provides visibility into how users interact with applications, information, and on-screen content, including proprietary and internally developed systems that third-party tools can't analyze. InnerActiv can capture and analyze on-screen application activity, detect misuse of specific application components, and surface behavioral patterns that indicate elevated risk, even when no data has moved and no external alert has fired.

‍

That means identifying things like:

  • Repeated navigation to high-value account types within an authorized portal
  • Unusual use of search functions or filters designed to surface sensitive records
  • On-screen viewing patterns that fall outside a user's normal responsibilities
  • Behavioral anomalies within proprietary systems that have no third-party integration
  • Information-gathering activity that precedes fraud but generates no traditional alert
    ‍

Rather than waiting for a suspicious transaction, organizations can find warning signs earlier and investigate behaviors that would otherwise go completely unnoticed.

‍

The TD Bank case is a reminder that many fraud schemes don't begin with malware or compromised credentials. They begin with trusted people, inside legitimate systems, doing things that only look suspicious once you can see what's actually on their screen.

‍

In many cases, the earliest indicators of fraud aren't in transaction records or security alerts. They're in the everyday on-screen actions happening inside applications long before the fraud itself is discovered.

‍

read next
In the News

Foreign Threats Don't Always Need Malware. They Need Trust.

November 10, 2026

Earlier this month, the CEO of a major tech company admitted to having a romantic relationship with a woman who U.S. officials believe was in the pay of a foreign government. The investigation is ongoing, but as of today, the company does not believe she was able to obtain sensitive information.

Company

InnerActiv Completes SOC 2 Type II Audit, Reinforcing Its Commitment to Security

November 5, 2026

Independent validation confirms that InnerActiv’s security controls operate effectively over time.

Company

InnerActiv Appoints Robert Schuett as Vice President, Customer Success & Operations

November 4, 2026

An insider risk veteran from Mandiant, DTEX, and the FBI will lead customer success, helping organizations safely adopt, use, and monitor AI.