Directive on Automated Decision-Making
Mandatory Treasury Board directive governing how Government of Canada departments use automated decision systems for administrative decisions. Requires an Algorithmic Impact Assessment before production use, plain-language notice and explanations to affected clients, bias testing, employee training, and human involvement scaled to a four-level impact rating; final decisions at the highest levels must be made by a human.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
Federal public servants, deputy heads, and program administrators involved in developing, procuring, or managing automated decision systems.
What the document contains
The directive covers the governance, design, procurement, and operation of automated decision systems used to make administrative decisions or related assessments about clients in Canadian federal departments. It details requirements for algorithmic impact assessments, transparency, explanations, testing, monitoring, data quality, security, and human involvement.
Not stated
- Automated decision systems used in federal administration must comply with administrative law and respect human rights.
- Departments must complete and publish an algorithmic impact assessment prior to bringing any automated decision system into production.
- Systems are categorized into four impact levels, which dictate specific requirements for notice, explanation, peer review, and human involvement.
- Departments must provide meaningful plain-language explanations and recourse options to clients affected by automated decisions.
- Ongoing quality assurance, testing for bias, data governance, and security measures are required throughout the lifecycle of the system.
· 1. Effective date · 2. Authorities · 3. Definitions · 4. Objectives and expected results · 5. Scope · 6. Requirements · 7. Roles and responsibilities of Treasury Board of Canada Secretariat · 8. Application · 9. References · 10. Enquiries · Appendix A - Definitions · Appendix B - Impact Assessment Levels · Appendix C - Impact Level Requirements
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Accountability
4/5Clear responsibility for AI outcomes
responding to cases of non-compliance in accordance with any Treasury Board instruments that address the management of compliance
s.8.3.3
screening score · 2026-08-19
Fairness
4/5Non-discrimination and equitable treatment
testing the data, information, and underlying model for accuracy, unintended biases and all factors that may unintentionally or unfairly impact the outcomes or violate human rights and freedoms
s.6.3.1
screening score · 2026-08-19
Human Dignity
2/5Respect for the inherent worth of persons
the equality, dignity, privacy, and autonomy of individuals
Appendix B, impact level descriptions
screening score · 2026-08-19
Human Oversight
5/5Humans remain in control of consequential decisions
The final decision must be made by a human. Decisions cannot be made without having clearly defined human involvement during the decision-making process.
Appendix C, Human involvement, levels III-IV
screening score · 2026-08-19
Privacy
3/5Protection of personal data and private life
made available to the public, while protecting privacy, security and intellectual property
s.4
screening score · 2026-08-19
Transparency
5/5Openness about how AI systems work and are used
Providing notice through all service delivery channels in use that the decision will be made or assisted by an automated decision system
s.6.2.1
screening score · 2026-08-19
Enforcement
4/5Consequences and compliance mechanisms with real force
responding to cases of non-compliance in accordance with any Treasury Board instruments that address the management of compliance
s.8.3.3
screening score · 2026-08-19
Impact Assessment
5/5Required assessment of an AI system's risks/impacts before or during use
Developing and maintaining the Algorithmic Impact Assessment tool and any supporting documentation
s.7.2; AIA required before production per s.6.1
screening score · 2026-08-19