AI irregularities cost large Canadian enterprises an estimated $144 million per year, with half those losses tied to governance failures rather than technology failures. That figure comes from a global IBM Institute for Business Value study surveying more than 1,000 senior leaders across 20 countries and 21 industries, released May 7 via IBM Canada’s newsroom.
The core finding: 63% of Canadian executives say gaps in AI governance already make it harder to deploy AI at scale. Only 18% of Canadian organizations say they currently have systems in place to coordinate and govern AI across everyday operations.
The Numbers
The $144 million annual cost breaks down into errors, bias, duplication, and uncoordinated deployments. The study attributes half of those losses to governance gaps specifically, not to failures of the underlying AI technology. The implication: organizations are losing money because they cannot see, audit, or control what their AI systems are doing, not because the systems themselves are broken.
“You can’t govern systems you can’t see,” said Manav Gupta, Vice President and CTO of IBM Canada, in the announcement. “AI systems now act as critical infrastructure, and that raises real questions about trust, accountability, and sovereignty for Canadian businesses, governments, and institutions.”
Digital Sovereignty as Operational Control
The study reframes digital sovereignty as operational control rather than data location. As AI agents enter healthcare, transportation, financial services, and public programs, organizations need four capabilities: visibility into which AI tools they use, understanding of how decisions are produced, control over access and updates, and the ability to intervene when systems behave unexpectedly.
“Digital sovereignty is about control, not isolation,” Gupta said. “Organizations need governance they can enforce and demonstrate, not just policies on paper.”
The Orchestration Gap
Organizations that coordinate AI governance across the full lifecycle (what the study calls “orchestration-led governance”) reported stronger productivity gains, higher returns on AI investment, and significantly fewer losses tied to AI errors. But the adoption rate is low: only 18% of Canadian organizations have reached that level.
The study was conducted jointly with the Dubai Future Foundation. IBM also announced IBM Sovereign Core, a software platform for building AI-ready sovereign environments, alongside the research.
Connection to Agent Governance
The findings land one day after Gartner published its inaugural Market Guide for Guardian Agents, which codified AI agent identity governance as a distinct enterprise security category. The convergence is clear: as autonomous agents move from pilot to production, governance is becoming the bottleneck, not capability. The IBM study quantifies what that bottleneck costs: $144 million per year for a large Canadian enterprise, with the problem concentrated in organizations that have no coordinated system for tracking what their AI does.