The following note is from our bi-weekly Re:Create Recap click here to subscribe.

Late last week, as Americans sleepily recovered from our turkey comas, a group of Canadian news media companies pounced, filing a copyright lawsuit against OpenAI on more or less the same theory as their US counterparts: training AI models with in-copyright data should require a license. Canada is no stranger to delayed-release copyright overreach, as Cory Doctorow recently highlighted: Canada enacted stupid DMCA anti-repair rules more than a decade after the US, at which point it was obvious the rules were harmful to the public. More recently, Canada extended its copyright term by 20 years, just like we did way back in 1998, as if we had learned nothing from the exercise. The Canadian publishers’ latest bandwagon gambit may have some edifying results, however, if it helps reveal important truths about copyright internationally:

1. Fair use is a key reason why the US is the envy of the world when it comes to innovation. It allows copyright to flex and accommodate new technologies and new cultural practices that couldn’t have been foreseen by congress. Like internet search before it, AI is a poster child for fair use.
2. At the same time, if the US wants to encourage other countries to be more hospitable to innovation and protect US companies from shakedowns abroad, the US gov’t should promote the adoption of US-style limitations and exceptions like fair use. The US Trade Representative (USTR) should abandon its policy of exporting only the exclusive rights in the law, and start touting the benefits of fair use. And if other jurisdictions want to compete in AI tech, they’ll welcome the opportunity for reform.
3. Many countries, including Canada, already have a version of fair use (aka fair dealing). Courts in Canada have interpreted the fair dealing provision quite broadly in recent years, such that it is arguably tantamount in scope to fair use. Fair dealing may be able to flex enough to foster AI development, if the Canadian courts continue to keep the public interest at the heart of their analysis.
4. Disputes about AI may further sharpen distinctions between different kinds of flexible exceptions in national copyright laws. The technology is so radically new that perhaps only the most flexible fair use-style exceptions will be able to accommodate it. This may prove to be a useful litmus test. Systems premised on natural rights/IP maximalism will struggle and sputter, while those based on a utilitarian/public good conception of copyright will comprehend this new technology more readily.