New AI rules are on the way

Nearly nine months after the release of ChatGPT, Canada is progressing toward increasing safety and transparency around generative AI. 

Driving the news: The federal government is currently pulling together a voluntary code of conduct that could commit firms to safety measures, testing, and disclosures, per The Logic.

  • Possible measures could include watermarking AI-generated content to differentiate it from human-created content and aggressively testing AI systems to find flaws. 

Why it matters: Canada’s proposed Artificial Intelligence and Data Act could take years to finalize and turn into law, but the code will warm up companies for incoming AI legislation. 

  • Amazon, Google, Microsoft, Meta, and OpenAI have signed on to a similar program in the US as the government prepares to sign AI legislation into law later this year.

  • The EU is also looking to pass a voluntary code of conduct as the bloc’s 27 countries work together to develop a universal set of laws—a process that could take years. 

Yes, but: Regulators are playing catch-up with a sector that’s racing to improve and update AI models, meaning that new rules could be outdated by the time they become enforceable.

Bottom line: With the plan in its consultation stage (and legislation a while away), it’s too soon to predict what AI regulation in Canada will look like and how effective it will be.—LA