Guide on the use of generative artificial intelligence

Summary

Guidance for federal institutions on generative AI risks (inaccuracy/hallucination, bias, privacy and security, IP, procedural fairness, over-reliance, undisclosed AI interactions, environmental costs) organised around the 'FASTER' principles (fair, accountable, secure, transparent, educated, relevant). It asks institutions to label AI-generated content, tell users when they interact with AI, test deployed tools for bias and vulnerabilities, including red teaming, plan independent audits, and route administrative-decision uses through the Directive on Automated Decision-Making.

Safety topics

Status notes

First released on 6 September 2023. The second version, of 20 February 2024, added enhanced definitions of the FASTER principles, sections on the responsibilities of federal institutions and on documentation requirements, and new frequently asked questions. A concise summary for public servants, Generative AI in Your Daily Work, followed on 15 October 2024. The guide says it will be updated regularly; its page was last modified on 24 September 2026. The Guide on the Use of Agentic Artificial Intelligence, published on 22 May 2026, complements it for agentic AI systems.

Sources

Open the official source https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/guide-use-generative-ai.html

Secondary source: https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/progress.html

Local copy

Verification

History

1. 2026-10-18verification.level: A → B — level computed from the evidence by tools/pipeline.py2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
2. 2026-10-18notes: First released 6 Sept 2023; second version 20 Feb 2024; the current page's metad → First released on 6 September 2023. The second version, of 20 February 2024, add — the content of the second version and the companion summary (15 Oct 2024) added from the progress page; the page-metadata issue date (2024-05-30), which is not in the page text, dropped; the agentic AI guide of 22 May 2026 noted as a complement2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
3. 2026-10-18safety_topics: ["public-sector use", "transparency & labelling", "accountability & liability", → ["public-sector use", "transparency & labelling", "accountability & liability", — 'evaluation & testing' added: the guide asks deploying institutions to test for bias and vulnerabilities, including red teaming, and to plan independent audits2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
4. 2026-10-18summary: Guidance for federal institutions on generative AI risks (inaccuracy/hallucinati → Guidance for federal institutions on generative AI risks (inaccuracy/hallucinati — environmental costs added to the risks, and the deployer practices of testing for bias and vulnerabilities (red teaming) and independent audits added from the guide's best practices2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)

Related records

DADM Canada · In force

GC AI Strategy 2025-2027 Canada · Published (non-binding, current)

Citation

Treasury Board of Canada Secretariat. Guide on the use of generative artificial intelligence. 2023-09-06. https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/guide-use-generative-ai.html. Status: Published (non-binding, current); verified 2026-10-18 (level B). Local copy archive/CA-04/2026-10-11/official.html retrieved 2026-10-11 13:59:34 UTC, SHA-256 a3e077c3d54856dfa0fcfbff492cc0eaad00963ac9d9d2090fbc3050ed7b0caa. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record CA-04.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.