Cullen International's latest Global Trends benchmark provides key insights on how 14 jurisdictions are addressing the growing intersection between copyright law and artificial intelligence (AI), focusing on:

• rules governing the use of copyright-protected material for AI model training;
   
• any mechanisms available to rights holders to opt out of such use (if by default, AI systems are allowed to use protected material for training purposes); and
   
• the conditions under which AI-generated content may qualify for copyright protection.

The research covers Australia, Brazil, Canada, China, the EU, India, Japan, Korea, Malaysia, Singapore, South Africa, Taiwan, the UK and the US.

According to the research, only three jurisdictions provide clear legal grounds allowing some form of AI training on copyrighted materials, while six jurisdictions are still debating the issue, and five currently prohibit or do not recognise such training under existing law.

Opt-out mechanisms are similarly uneven: only the EU has a defined, machine-readable opt-out requirement; two jurisdictions have proposals under legislative debate; and the remaining 11 jurisdictions do not foresee any formal mechanism.

On copyright protection for AI-generated content, only three jurisdictions explicitly recognise protection of the output where a human contribution can be identified.

For more information and to access the full benchmark, please click on “Access the full content” – or on “Request Access”, in case you are not subscribed to our Global Trends service.

See also our detailed research on related initiatives in the Americas and in Europe.