Governing AI: A Plan for Canada
White paper by AI Governance and Safety Canada (AIGS) outlining five high-impact actions for the Canadian government to significantly advance AI governance by end of Q2 2024. The paper covers the evolving AI risk landscape from narrow AI through to potential AGI, key considerations for AI governance, and concrete recommendations for Canada.
Organizations
- AIGS — publisher
Links
Policy recommendations
- rec-2023-001 Establish a central AI agency for the Government of CanadaUntracked
Create an expert team based within the Privy Council Office to monitor and inform leadership about AI developments, communicate with the provinces, and coordinate federal action. ISED and other departments lack the institutional incentives or mandate to navigate and balance the full range of issues.
- rec-2023-002 Invest >$500M in AI governance and safety researchUntracked
More than double the existing CIFAR Pan-Canadian AI Strategy funding, earmarking the extra funds to developing better policy and technical AI safety solutions. Governance and safety research must rapidly catch up to the hundreds of billions invested annually in capabilities.
- rec-2023-003 Champion and fund global AI governance effortsUntracked
Support global initiatives such as at the UN, G7, UK safety summit, and OECD. Lead the conversation about CERN, IPCC, and IAEA models for global AI governance. Without global coordination, domestic regulation will simply push AI development to other jurisdictions.
- rec-2023-004 Improve and pass the AI & Data ActUntracked
Accelerate consultations and drafting of legislation as the best mechanism for enacting policies on audits, transparency, and liability. Establish a Canadian AI Commission to regulate high-risk models.
- rec-2023-005 Launch a national conversation on AIUntracked
Hold nationwide open public consultations to inform policy decisions on AI in relation to the jobs transition, wealth concentration, and the right balance between pursuing rewards and limiting risks. Canada's relative stability and strong AI ecosystem make it well placed to lead this conversation.