Obligations on Vendors to Disclose Whether and How They Use AI: A Supervisory Expectation for Regulated Industries in Canada and the UK
As you navigate the market for artificial intelligence (AI) solutions, a new expectation has emerged from the Office of the Superintendent of Financial Institutions (OSFI): vendors must clearly disclose how they use AI, creating an unprecedented level of transparency in this space. This change demands immediate attention from regulated organisations across healthcare, insurance, and financial services, as their exposure to capability risk, governance risk, and liability risk increases significantly.
AI solutions are no longer viewed as mere tools; rather, they are complex systems that can impact decision-making processes and have direct implications for patient care, regulatory compliance, and customer trust. Without a clear understanding of how AI is used, organisations face the risk of deploying inaccurate or biased technology, making it challenging to defend their decisions in front of regulatory bodies.
OSFI’s expectation poses significant challenges for vendors selling AI solutions in Canada and the UK. The regulator now expects these companies to provide detailed information about their use of AI, including data on how they collect and handle user information, as well as the transparency of their algorithms’ decision-making processes. This increased transparency will allow organisations to make more informed purchasing decisions but also increases their exposure to potential failures or inaccuracies in the AI systems.
To mitigate this risk, OSFI advises organisations to carefully evaluate the capabilities of any proposed solution against real operations and governance obligations before committing resources or patient trust. This includes assessing whether the technology does what was claimed, can be defended by corporate governance when regulators ask questions, and what liability is incurred when errors occur. Straven & Co stands ready to help organisations like yours make informed decisions about AI solutions by examining the proposed solution against real operations, governance obligations, and regulatory exposure.