AI Laws & Regulation: USA, Canada, and Europe

I did some digging into emerging laws and regulations covering emerging AI tech. Turns out there’s a lot going on!

As with any new technology, there’s a mix of perceived risk and reward, along with efforts to figure out what exactly is “new” about AI (ie: what is or isn’t covered by pre-existing laws and regulations).

Much of the focus is on “high-risk” sectors or applications (more on what these are below), so present discussions have little relevance to us at REW right now. This is because we’re focusing on the “easy wins” side of the AI equation, specifically: providing high-value services quickly and efficiently, and increasing access to information and services. Many emerging regulations are actively concerned with leaving such uses “as is”.

However, it’s still good to keep an eye on the emerging regulatory landscape for a few reasons:

  1. Laws and regulations can have secondary effects or unintended consequences that might effect us.
  2. Seeing the values and concerns embedded in these laws will help us focus our oversight efforts.

So here’s a region-based rundown of how things are going…

Europe has arguably made the quickest and strongest moves towards regulating the tech sector, including AI, and they’ve developed a reputation for coming down pretty hard.

Europe was also ahead of the curve for a while, issuing a statement on AI as far back as 2018. However, the problem is that those early statements and commitments now seem (surprise surprise) a little out of date.

For years the big worries about AI have been about very specific applications like social credit systems, facial recognition, media manipulation (deepfakes), or bias in systems that control access to certain opportunities (financial services, housing, education, employment etc.). These cover some of the afore mentioned “high risk” applications of AI.

However, LLMs (ChatGPT etc.) present a bit of a problem for these frameworks because of how general they are. As such, the EU is wary about introducing any regulations which unfairly limit the low-risk or high-value uses of this tech, while also being concerned about where such general-purpose capabilities could go.

Unlike the strong centralized approach of the EU, the US Government’s strategy favors deferring regulation to specific agencies or regional governments. Laws and regulations are already popping up at the state and even city level.

So there doesn’t seem to be any move to regulate AI directly, just commitments to maintain current laws and regulations (regarding discrimination, for example) in relation to AI. And there have been a bunch of statements to this end, including the AI Bill of Rights and Joint Statement on the Enforcement Efforts etc. etc. etc.

And here’s a pretty good summary comparing the differences and similarities btw the US and Europe on AI regulation.

Ultimately, it seems that AI regulation in the US will be a bit of a patchwork.

Unlike the U.S., Canada seems OK with introducing legislation at the federal level. For example, in June 2022, Canada introduced the Artificial Intelligence and Data Act (AIDA? They couldn’t find one more letter?) which outlined where specific legislation and regulation would be going. However, unlike the EU’s heavy-handed approach of proposing outright bans on certain uses or systems, Canada will instead be identifying “high impact AI systems” and then applying certain risk-mitigation measures, like requiring companies to submit plans for oversight and transparency.

As mentioned, none of REW’s work with AI is likely to be directly affected. Real estate is, of course, a quite highly regulated sector. But since our goals, especially in our Global MLS project, are to create helpful content and enhance accessibility with AI, we’re pretty confident of being on the right side of any coming legislation.

One caveat is this: There WILL be legislation. In the next few years we’ll be seeing a lot of laws and regulations being tabled and coming into effect. Some might even be poorly crafted and vague in intent or application (:exploding_head:).

All of this just re-affirms the need for close human oversight and expertise in the use of AI technology. That’s why REW is committed to the concept of human-assisted AI for real estate applications, not just to make sure that we’re always compliant with relevant regulations, but to make sure we’re consistent with our own values and responsibilities as well.


Thanks for looking into this and providing the summary! Interesting.