Skip to content

Content Library

Back to Articles

Article August 6, 2026

Why Defensible Controls Will Define AI Risk Success

Why Defensible Controls Will Define AI Risk Success

Financial institutions should focus less on deploying artificial intelligence and more on creating integrated, defensible control frameworks that can withstand regulatory scrutiny, according to a technology executive, who argued that explainability, gover



  • Financial institutions should build integrated control frameworks that connect policies, monitoring, investigations and remediation rather than relying on fragmented systems
  • Explainable AI, traceability and reproducible evidence chains are becoming essential to regulatory confidence
  • Technology alone cannot deliver defensible controls without governance, ownership and continuous improvement
  • Purpose-built AI models should support qualified professionals rather than replace human judgment
  • Organizations that unify data, controls and feedback loops will improve resilience, accelerate investigations and strengthen regulatory readiness



Artificial intelligence will only strengthen risk management if financial institutions build integrated control frameworks that connect policy, surveillance, investigations and remediation into a single defensible operating model, a senior technology executive has claimed.

Rather than viewing AI as a standalone solution, organizations should concentrate on creating end-to-end governance that enables them to explain decisions, demonstrate effective controls and respond rapidly to regulatory scrutiny.

Speaking at CeFPro's 2026 Risk Evolve conference, the Global Head of Account Management and Customer Success at a global compliance technology provider said many organizations continue to struggle with fragmented controls spread across multiple systems and departments.

While most firms have established policies, monitoring capabilities, approval processes and case management functions, these often operate independently, making it difficult to demonstrate how risks are identified, managed and continuously improved.

"When your firm faces an examination or investigation, how long does it take to reconstruct a full evidence chain?" the executive asked delegates.

She argued that many institutions still spend days or even weeks assembling information from disconnected systems, creating inconsistent outcomes, weak evidence trails and delayed remediation when regulators request supporting documentation.

She added that truly defensible controls depend on four fundamental principles: traceability, reproducibility, governance and scalability.

Every regulatory obligation, she said, should map directly to specific preventive and detective controls.

Further, she believes. every investigation should be reproducible from beginning to end, every control change should be governed and documented, and every framework should operate consistently across jurisdictions, business lines and communication channels.

"Tooling alone is not the answer," she said.

While technology can dramatically improve consistency and efficiency, she stressed that successful control environments ultimately depend upon organizational discipline, clearly defined ownership and a culture of continuous improvement rather than software alone.

The executive outlined a six-stage lifecycle for creating what she described as a defensibility-grade controls framework.

Organizations should first identify regulatory obligations and internal policies before designing preventive and detective controls, monitoring activity, investigating incidents, implementing corrective actions and feeding lessons learned back into the overall framework.

Closing that feedback loop, she argued, enables firms to adapt as regulations, technologies and business risks evolve.

She illustrated the importance of effective remediation by recalling a customer that experienced a sudden increase in outbound data transfers shortly after employees shifted to remote working during the pandemic.

Initial concerns centered on potential data theft. However, further investigation revealed that many employees were forwarding meeting invitations to personal devices simply because their corporate laptops lacked webcams.

Rather than disciplining staff, the organization resolved the issue by issuing webcams across the workforce, causing alerts to decline significantly while addressing the underlying operational problem.

The discussion also focused heavily on artificial intelligence. While enthusiasm surrounding large language models continues to grow, the executive warned that organizations should carefully evaluate how AI models are trained, validated and governed before relying upon them for surveillance or compliance activities.

"The worst thing I can have a customer say to me is that AI is a black box," she said. "It is not a black box. It is explainable."

She argued that organizations should expect vendors to demonstrate not only how AI models generate recommendations but also how they have been independently tested for accuracy, consistency and reliability.

Events you may be interested in

View all of our upcoming events

Sign Up to Our Newsletter

Sign up

Connect Plus

Unlock more with Connect Plus membership

This content is available to Connect Plus members.