Banks deploying artificial intelligence across operations are confronting a parallel problem: the same technology accelerating banking automation is also giving attackers faster and more adaptive ways to target financial infrastructure. Finzly is responding by introducing Assure, an AI-powered security, compliance and operational resilience layer for its BankOS platform, designed to continuously monitor threats, controls and compliance signals rather than relying primarily on periodic security reviews. The launch adds another dimension to the emerging agentic banking model, where AI is increasingly being positioned not only to perform banking work, but also to protect the systems and autonomous processes doing it.
Finzly brings AI into the security layer
Finzly said Assure will form part of its Agentic Galaxy suite of AI capabilities and will be integrated into BankOS, the company’s operating system for banks.
Rather than deploying separate AI security tools after banking applications have been implemented, the company is positioning Assure as an intelligence and assurance layer embedded within the infrastructure supporting banking operations.
The approach is intended to give financial institutions continuous visibility across security, compliance and operational resilience while they introduce new digital services and AI-powered workflows.
According to Finzly, Assure can monitor signals across the technology environment, identify anomalies, test controls and flag potential weaknesses. AI is also being used to expand the amount of assurance activity that can be performed without requiring equivalent increases in manual security and compliance work.
Finzly founder and CEO Booshan Rengachari said banking is entering an environment where attacks can increasingly occur at machine speed, requiring security and intelligence to develop alongside AI-led innovation.
The company currently retains human oversight as it develops and tunes Assure’s processes, while more automated responses are planned as the technology matures.
AI changes the speed of the security contest
The launch reflects a broader shift in cybersecurity as artificial intelligence becomes available to both financial institutions and attackers.
Finzly cited TrendAI’s Modern Bank Heists 2026 report, which found AI-enabled attacks against financial institutions increased 89% year over year. The research also reported that 67% of institutions had encountered attackers actively attempting to counter defensive measures during live incidents.
That creates a different security environment from one dominated by predetermined attacks and manual intervention. AI-enabled systems can potentially adjust techniques, automate reconnaissance and coordinate multiple stages of an attack considerably faster than traditional processes.
For banks, this increases pressure to shorten the time between identifying unusual activity and understanding whether it represents a genuine threat.
Assure is designed to use AI to accelerate investigation and triage, continuously monitor attack-surface risks and support security testing across both conventional and AI-assisted software development.
The platform will also collect compliance evidence continuously, reducing reliance on teams manually assembling information when controls need to be demonstrated.
Continuous assurance could change compliance monitoring
One of the more significant aspects of the announcement is Finzly’s emphasis on continuous assurance.
Bank security and regulatory controls have traditionally involved substantial amounts of periodic testing, documentation and evidence gathering. While those processes remain necessary, increasingly dynamic technology environments can make point-in-time assessments less representative of what is happening between reviews.
AI creates the possibility of monitoring more controls continuously.
A system could identify changes in security signals, detect unexpected behaviour and verify whether particular controls continue to operate as intended. Exceptions could then be escalated to specialists rather than requiring people to perform every stage of the monitoring process manually.
For financial institutions managing increasingly complex cloud, API and AI environments, that could allow assurance processes to operate closer to the speed at which the underlying infrastructure changes.
Finzly Chief Information Security Officer Supro Ghose said security needs to evolve alongside the adoption of AI across financial institutions, with AI itself becoming part of the defensive response to AI-enabled threats.
Agentic banking creates new security questions
Assure also extends the Agentic Galaxy strategy Finzly introduced in October 2025. The company initially positioned Agentic Galaxy around AI capabilities for payment operations and banking workflows and has since expanded its use across operations, software development, testing and customer experiences.
As those systems become more autonomous, the security challenge changes.
Traditional applications generally operate according to predefined instructions. AI agents can potentially interpret information, make decisions and initiate actions across several systems. This makes identity, permissions, monitoring and auditability increasingly important because banks need to understand not only whether a system has been compromised, but also whether an authorised agent is behaving as expected.
The distinction will become more important as financial institutions move from AI assistants towards agents capable of completing operational tasks.
Security architecture may therefore need to evolve alongside agentic architecture. The same intelligence that allows banking systems to operate more autonomously could increasingly be required to monitor those systems continuously and identify behaviour that falls outside expected boundaries.
AI security moves closer to banking infrastructure
Finzly’s approach also illustrates how AI is moving deeper into the banking technology stack.
Early deployments frequently treated AI as an application sitting above existing systems. The next stage is increasingly about embedding intelligence into infrastructure, operational workflows, software development and security controls.
For banks, this could reduce the distinction between an AI strategy and a cybersecurity strategy. Institutions introducing autonomous systems will need controls capable of operating at comparable speed, while security teams may increasingly depend on AI to process the volume of signals generated across complex technology environments.
The competitive question for banking technology providers may consequently expand beyond how effectively their platforms enable AI. Banks could also begin examining how those platforms govern, monitor and defend the intelligent systems operating within them.
What it means for the industry
- AI is becoming both an attack and defence technology: As attackers use AI to increase speed and adaptability, banks are likely to rely more heavily on AI-driven detection, investigation and response.
- Continuous assurance could supplement periodic controls: Automated monitoring may allow banks to identify control failures and emerging risks between traditional assessment cycles.
- Agentic AI expands the security perimeter: Autonomous agents introduce new questions around identity, permissions, behaviour monitoring and accountability.
- Human oversight remains important: Finzly is retaining people within the assurance process while its technology develops, illustrating the cautious approach likely to accompany greater security automation.
- AI and cybersecurity strategies are converging: As intelligence becomes embedded deeper within banking infrastructure, institutions will increasingly need to develop AI capability and AI security together.
Article Source: Finzly
