**Canada’s AI Safety Institute and Safe and Secure AI Advisory Group: A Step towards Clearer Evaluation Language for Regulated Industries**
The Canadian AI Safety Institute and the UK-based Safe and Secure AI Advisory Group aim to create a clearer understanding of AI evaluation language in Canada, promoting safety in regulated industries like healthcare, insurance, and financial services. For organisations deploying or acquiring AI solutions in these sectors, this development brings attention to their potential exposure through capability risk, governance risk, and liability risk.
The Canadian AI Safety Institute aims to normalise the use of a common evaluation language for AI systems, allowing users to more effectively assess AI capabilities and make informed decisions about implementation. This effort could help reduce the likelihood of deploying systems that do not meet claims or cannot be defended when regulatory bodies ask questions. It may also lower liability risks associated with AI failures.
Organisations must take action now by examining their current AI solutions for capability risk, governance risk, and liability risk. They should assess whether vendors’ claims align with real-world operations, ensure that decision-making processes can withstand scrutiny from regulatory bodies, and consider the potential consequences of AI errors or failures.
To mitigate these risks, a prudent organisation would take clear steps in a reasonable timeframe. First, they would re-evaluate their AI solutions against the latest capabilities and vendor claims to identify any discrepancies. Second, they would reassess their decision-making processes to ensure alignment with regulatory requirements and internal governance frameworks. Third, they might engage external experts to conduct a thorough examination of AI systems before deployment.
Straven & Co is an independent AI validation firm that provides unbiased examination of AI solutions against real-world operations, people, governance obligations, and legal and regulatory exposure, delivering a plain verdict: proceed, proceed with conditions, or do not. As an expert in AI evaluation language, Straven helps organisations navigate the complexities associated with AI deployment by offering a trusted and independent view on AI capabilities.
Straven & Co examines AI decisions before they are acted on: stravenandco.com