Glossary · Canada
Canadian Voluntary Code of Conduct on Generative AI
Also known as: Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems, Canada’s generative AI code
The Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems is a code published by the Government of Canada in September 2023, in which signatory organizations commit to measures for accountability, safety, fairness, transparency, human oversight and robustness when developing or managing advanced generative AI systems.
- AI regulation
- Canada
In one sentence
Canada’s voluntary code (2023) commits signatories to safety, fairness, transparency and oversight measures for advanced generative AI.
Example
A signatory publishes a description of its model’s capabilities and known limitations and offers a tool to detect content generated by it.
How it applies
- Measures: Grouped under six principles: accountability, safety, fairness and equity, transparency, human oversight and monitoring, and validity and robustness. Transparency measures include publishing information on the capabilities and limitations of systems and providing a way to detect AI-generated content, such as watermarking.
- Status: Voluntary and signed by dozens of Canadian and international organizations. Without AIDA, it remains Canada's main federal instrument addressed to private AI developers.
- Technical documentation: Publishing capabilities and limitations is a documentation task — similar in content to model cards and to the information the EU AI Act requires for general-purpose AI models.
Voluntary code vs. the EU AI Act
The code asks for measures similar to the AI Act's GPAI obligations and its GPAI Code of Practice, but there is no law behind it: no regulator can require signatories to comply.