Christopher Perrin was recently featured on BBC News to comment on the legal implications of OpenAI’s new video generation model, Sora 2. The segment aired during both the 6pm and 10pm bulletins and explored the copyright concerns emerging from this latest AI development.
Christopher broke down some of the key issues surrounding Sora 2, which include:
- Opt-Out Backlash: Sora 2 launched in the US with an opt-out approach - allowing users to generate content featuring copyrighted characters (like Pokémon or Mario) unless rights holders explicitly objected.
- Legal & PR Fallout: This triggered a swift backlash from creators and companies like Nintendo, who pointed out that copyright laws in most jurisdictions don’t support this model.
- OpenAI’s Response: CEO Sam Altman responded swiftly, announcing a shift to an opt-in model, hinting at future revenue-sharing arrangements - but this is still evolving and far from watertight.
- Legal Grey Zones: There’s a fine line between training AI on copyrighted data (which may be “fair use” and permitted in some places) and replicating protected characters or styles - which is more likely to be infringement. Many critics argue Sora 2’s outputs appear to have crossed that line. It’s fair to say that this is a bit of a legal minefield, as AI law is still developing, albeit at a slower pace than the technology is moving.
- Jurisdictional Complexity: AI laws vary globally. The UK, for instance, has no direct AI legislation yet - but copyright (and various other) laws still apply. What’s legal in Japan, for example, may not be in the UK or EU.
- Global Harmonisation Needed: There’s a growing argument in favour of harmonisation of AI laws globally, although that’s clearly not an easy thing to do. Whilst there is already a Council of Europe AI Treaty, signed in 2024 by EU countries and several non-EU states such as the UK, US, and Japan, which sets out broad principles for ensuring that artificial intelligence respects human rights, democracy, and the rule of law, it is not sufficiently detailed to address all of the complex legal and ethical issues raised by AI. For example, while the treaty acknowledges that transparency and accountability must be balanced with the protection of intellectual property rights, it does not provide specific guidance on how to handle copyright or data-training questions. Given that copyright laws differ across jurisdictions and AI technologies are deployed globally, this is problematic.
As AI capabilities accelerate, Sora 2 is a powerful reminder of the urgent need for clearer, harmonised global standards.
Further information
If you have any questions regarding this blog, please contact Christopher Perrin in our Corporate, Commercial & Finance team.
About the author
Christopher Perrin is a highly experienced solicitor who leads the Corporate, Commercial and Finance team’s general Commercial & Technology Contracts, Outsourcing & Data legal advisory services.
Share insightLinkedIn X Facebook Email to a friend Print