Industry and Technology

The House of Commons Industry and Technology committee met to study artificial intelligence, hearing from Sean Mullin, a senior fellow at the Munk School of Global Affairs and Public Policy; Jennifer Brown, CEO of the Society of Composers, Authors and Music Publishers of Canada; and Michael Lee, chief strategy officer of UrbanLogiq.

Sean Mullin argued that Canada’s AI sovereignty should aim for freedom from coercion, not isolation, and that the country’s most acute controllable vulnerability is cloud infrastructure. He recommended a tiered approach to sovereign data centres, prioritizing national security and sensitive data, and said decisions made now about infrastructure and procurement will harden into long-term commitments. He noted that interdependence with like-minded middle powers is not the same as dependency on a dominant power, and that Canada has a window of five to 10 years to act.

Jennifer Brown said AI companies are using copyrighted music without permission or compensation, posing an existential threat to creators, and that up to 24% of music creators’ revenues could disappear by 2028 under current conditions. She recommended that Canada reject a text and data mining exception to the Copyright Act, citing Australia and the U.K. as examples, and establish transparency obligations requiring AI companies to disclose training data and label AI-generated outputs. She argued that opt-out systems are unfair because they put the onus on creators, and that copyright protection should remain tied to human creativity.

Michael Lee said the most valuable work in government AI deployments is integrating existing data, not building models, and that successful deployments require audit trails and workforce plans. He recommended investing in data integration before AI models, mandating auditability standards as procurement criteria, requiring workforce plans for government AI contracts, and mandating structural AI governance with risk-calibrated controls. He emphasized that the committee should give equal attention to the cost of moving too slowly on responsible AI adoption, distinguishing between proven, governed systems and frontier systems that require more caution.

In response to questions, Mullin said that unrestricted frontier AI models would be irresponsible and that regulations should balance innovation with mitigating harms to national security, privacy and catastrophic risk, while also noting that the U.S. administration has sought leverage against Canada. Brown confirmed that the cultural sector’s requests align with the Coalition for the Diversity of Cultural Expressions, and that transparency is the first principle needed to enable licensing markets. Lee said UrbanLogiq uses strict privacy standards including audit trails, access controls, encryption and zero data retention, and that government should use these tools to address public safety risks like wildfires and geohazards.