EU AI Act High-Risk Obligations Reach Full Enforcement: What This Means for Healthcare and Financial Services in Canada and the UK
As of August 2026, all buyers of high-risk artificial intelligence (AI) will be forced to demonstrate evidence-based risk management, data quality, and human oversight. For healthcare, insurance, and financial services companies in Canada and the UK, this development poses significant challenges and opportunities.
Risk arises from three areas: capability, governance, and liability. Capability risk is about whether high-risk AI does what it claims to do; governance risk revolves around defending AI decisions when asked by regulatory bodies or courts; and liability risk focuses on accountability for AI errors. In a rapidly evolving regulatory environment, organisations must ensure their AI systems operate in line with the EU AI Act’s new requirements.
To mitigate these risks, prudent companies should take immediate action. This includes:
1. Conducting a thorough review of current AI investments to identify high-risk solutions.
2. Developing clear data quality and management procedures for all AI projects.
3. Implementing robust human oversight mechanisms for high-risk AI systems.
In practice, many organisations struggle with these tasks due to vendor information asymmetry, where vendors have greater insight into the product’s capabilities than buyers can verify themselves. Internal teams may also be too close to the decision-making process to deliver objective assessments. As a result, organisations often find it difficult to test claims or evaluate AI systems independently.
Straven & Co offers independent validation services that bridge this gap by examining an organisation’s specific AI decision against its real operations, people, governance and legal exposure, and vendor claims. Our plain verdict – proceed, proceed with conditions, or do not – provides the necessary confidence for companies to move forward securely in the rapidly evolving regulatory landscape.
Straven & Co examines AI decisions before they are acted on: stravenandco.com