How attackers target banks through trusted logins and tools

What Elastic Security’s November 2026 threat report means for financial institutions and the teams that protect their customers, systems, and payments

A user is presented with a pop-up or CAPTCHA that looks like a routine verification step when browsing to a seemingly legitimate website. It asks them to copy and paste a command. They follow the instructions, and the command runs on their device. By the time the customer opens their banking app or signs into a financial website, an attacker may already have a foothold on the device they use to manage their money.

Elastic Security’s’ November 2026 threats, etc report highlights two regional banking fraud toolkits in its coverage of paste-and-run attacks. SCMBANKER was part of an operation targeting customers of Mexican banks, fintechs, payment processors, and cryptocurrency exchanges. TCLBANKER, a Brazilian banking Trojan, monitored browser activity for targeted financial sites and could display fraudulent full-screen overlays. The research on TCLBANKER also documents a tampered installer and modules that spread through WhatsApp and Outlook. 

These campaigns show one way an attacker can exploit a familiar interaction. Attackers used a genuine Microsoft sign-in page, approved AI applications, and legitimate remote management software to gain or maintain access. Each looks familiar in isolation. What matters is what happened next. Read the Elastic Security Labs overview.

Follow the session past sign-in

In one incident, a phishing kit sent the victim to Microsoft’s genuine device sign-in page. The victim entered a code and approved the request. That approval issued a token to the session initiated by the phishing kit. Elastic subsequently observed successful Microsoft Graph API calls and the registration of a virtual device. The sign-in was one step in an attack that continued across identity and cloud activity. See the identity investigation in the threat report.

Consider a treasury team. The security operations center (SOC) sees an unexpected device registration tied to a treasury employee. The payments team sees a request to change a beneficiary. Either event might have a routine explanation. Together, they give investigators a reason to check who initiated the change, how it was approved, and whether a payment is pending.

That example is a useful test of the handoff. How quickly can the identity team establish what the session accessed? Can the payments team independently verify the instruction? Who can hold a pending payment while the two teams compare what they know?

Monitor what approved software does after it starts

Elastic documented two intrusions in which sanctioned AI applications were used to run commands. 

  • On Windows, an operator used Claude Code to stage tooling and reach an external GitLab instance through a proxy tunnel. 

  • On macOS, the ChatGPT desktop application acted as a shell command proxy in an intrusion involving credential access. 

In both cases, monitoring which application launched each process helped reveal the activity. See the AI-assisted offense section.

For a financial institution rolling out AI coding assistants, the practical question is what developers expect those tools to do. Launching a shell or downloading a dependency may be routine. Reading a credential store and then opening a tunnel to unfamiliar infrastructure is a sequence worth investigating. The SOC needs the endpoint and network context to distinguish the two, and engineering teams can help establish what normal use looks like.

The same question applies to remote support. The report describes attackers using legitimate remote monitoring and management software for interactive access. For a bank that relies on vendors to support critical systems, the software’s name is only one part of the investigation. The SOC also needs to know who initiated the session, which system it reached, and whether its activity matched the work the vendor was authorized to perform. See the remote management findings.

Give security and fraud teams a way to act on each other’s signals

The banking malware cases highlight an important visibility gap. If malware runs on a customer’s device, the financial institution may first see the effects in its own channels, such as an unfamiliar session, a new payee, or a transfer that differs from the customer’s usual activity. When the compromised device belongs to an employee or vendor, the SOC may see the technical evidence first. In either case, the people who can investigate the compromise and the people who can intervene in a transaction may sit in different teams.

A cross-team exercise can highlight this visibility and response gap and provide opportunities to address it before you’re in an incident. Give the SOC a suspicious session and give fraud operations a pending payment following a beneficiary change. Ask each team to work from the information it normally receives. Can they match the session to the business activity? Can they verify the instruction through an independent channel? Does someone have the authority to pause the transaction while the investigation is underway?

The same exercise works for an insurer reviewing a change to claims disbursement details or a capital markets firm investigating revised settlement instructions. Use a workflow where timing matters and measure how long it takes to connect the events, reach the decision-maker, and act.

Measure response against the business outcome

Elastic Security’s threat report gives defenders concrete techniques to hunt for from suspicious commands launched by an AI assistant to unexpected device registration after a sign-in. The observations describe activity seen in Elastic’s research and telemetry during the reporting window. Financial institutions can use those techniques to test their own controls and response procedures. See the report methodology.

A practical starting point is to pick one sensitive workflow and trace it from sign-in through the final business action. Identify what the SOC, identity team, and fraud or operations team can each see. Then run the exercise and ask the question that matters when money is about to move. How much time passes between the first suspicious event and the moment someone can intervene?

Read the full November 2026 threats, etc report for the attack sequences, detection guidance, and related research.

Learn more

Get in touch to learn more about how Elastic can support your fraud prevention and security teams.

The release and timing of any features or functionality described in this post remain at Elastic's sole discretion. Any features or functionality not currently available may not be delivered on time or at all.

In this blog post, we may have used or referred to third party generative AI tools, which are owned and operated by their respective owners. Elastic does not have any control over the third party tools and we have no responsibility or liability for their content, operation or use, nor for any loss or damage that may arise from your use of such tools. Please exercise caution when using AI tools with personal, sensitive or confidential information. Any data you submit may be used for AI training or other purposes. There is no guarantee that information you provide will be kept secure or confidential. You should familiarize yourself with the privacy practices and terms of use of any generative AI tools prior to use. 

Elastic, Elasticsearch, and associated marks are trademarks, logos or registered trademarks of elasticsearch B.V. in the United States and other countries. All other company and product names are trademarks, logos or registered trademarks of their respective owners.