The U.S. has taken the strictest stance to date on artificial intelligence laws compared to the international community. U.S. courts and the U.S. Copyright Office have determined that only human authored creative works are eligible for copyright protection. Any works created by AI would not qualify for such protections. The U.S. Copyright Office has even gone so far as to require copyright applicants to disclose what portions of the work were created by AI.  

Meanwhile, China's courts have swung the other way - allowing copyright protection for AI generated images. In the U.K., an old law called the Copyright, Designs, and Patents Act of 1988 provides for protection of computer-generated works. It was initially drafted to guard against the concern that any work that used a computer would lose protection under copyright law. Of course, with the advent of AI, this law is being reexamined and will likely be re-litigated to determine how generative AI fits in and whether those works may be protected under this 1988 law.

The U.S.'s stricter approach - made even stricter by the requirement of registration prior to suing to enforce a copyright right - may lead to stagnation of AI generated works and, by result, be less welcoming to the businesses that develop AI technologies.

The other major issue regulators are trying to sift through is whether it is permissible to use copyrightable works to train AI. In the U.S. the issue revolves around whether such use is a “fair use” and therefore would not require permission from the copyright owner to use. The U.S. cases on this issue are still in their early stages and a similar case in the U.K. has just been greenlit to proceed to trial. 

Meanwhile, Japan has amended its copyright laws to allow data scraping for training machine learning models, generally making Japan friendly to AI technology companies. Singapore followed suit. Israel has determined such training would be a fair use and therefore permissible, though they included an exception that training exclusively on the work of one artist to compete with that artist would not be permissible. The EU's draft of the upcoming AI Act adopts EU copyright law and allows exceptions for data scraping, but allows for opt-out provisions for rights owners. 

It is unclear how all of these laws will interact with one another in this global economy. What will happen when an AI program trained in a country with friendlier copyright law is then used or commercialized in a jurisdiction with harsher laws? It is murky at the moment, but technology companies are watching as these laws are written and legal cases are decided. The jurisdictions that will be more supportive of AI are likely to see a boost in creative output and economic growth due to technology companies seeking a safe haven for their work.