If we have selected the wrong experience for you, please change it above.
Can the watchdog use the tool it must also question?
GenAI and the second line
GenAI presents a dilemma for the second line. Its role is to provide independent, sceptical oversight by challenging first-line methods, testing controls, and assuring the board and regulator that risks are understood. When that function relies on technology that is hard to interrogate, questions of dependency and defendability arise. This article describes where GenAI can strengthen second line oversight.
Fragmented monitoring is the problem, not slow paperwork
Many second line functions suffer less from capacity constraints and more from fragmentation. Oversight is dispersed across domains, teams and jurisdictions that apply different standards and see only parts of the picture. The result is inconsistent risk ratings, competing regulatory readings and duplicated assessments that cannot be reconciled into a firm‑wide view. That fragmentation is the weakness GenAI is uniquely able to address: not by simply working faster, but by making oversight coherent.
How GenAI strengthens oversight
GenAI delivers three high‑value outcomes for the second line.
Examples from current industry practice
Common applications across financial services include:
The independence question: three non-negotiable conditions
The benefits described above are only usable if the second line can stand behind the tool that produces them. That requires treating GenAI with the same rigour applied to any method used for assurance.
Model validation and monitoring must be formal and fully documented. Teams should record a validation plan, agreed performance metrics such as accuracy and false‑positive rate, procedures for detecting drift and a schedule for periodic revalidation.
Explainability and traceability are essential. Every output should be linkable to the source data and to the chain of reasoning so auditors and regulators can follow how conclusions were reached.
Finally, the second line must apply the same governance it expects from the business: document decisions, assign a named accountable owner for each model and run continuous monitoring for drift and performance degradation.
Handled this way, GenAI does not dilute independence. It can strengthen it by enabling broader, more consistent oversight that is defendable under scrutiny.
What changes for the people in the function
Adopting GenAI changes how expertise is held and used. By codifying established methods into tools, GenAI reduces concentration risk among a small group of specialists and makes guidance accessible to less experienced staff, shortening onboarding time. Automating routine monitoring and reporting lets senior specialists focus on interpretation, challenge and strategy, the oversight work that cannot be delegated to a model. Organisations will need capabilities in model governance, explainability and data engineering, and must assign clear accountability for material decisions.
The case for acting
For the second line, the primary return on GenAI is credibility rather than efficiency. When implemented with appropriate governance, GenAI can produce a coherent firm wide view of non-financial risk, enable defensible and documented regulatory interpretations, surface systemic issues earlier and reduce reliance on a handful of irreplaceable experts. Organisations that succeed do not merely automate existing tasks. They use the technology to build a single, defendable risk narrative and put in place the controls and evidence needed for the second line to explain exactly how its conclusions were reached.
