Who owns the intellectual property in AI Generated Images (USA Edition)?

23rd August 2023

I wrote back in May that people kept asking me the same question: “Hey Raoul, who owns the intellectual property in images or art created by generative AI?“.

Guess what? Someone has finally asked a US District Judge a similar thing. So now we know what the USA Legal System’s answer to that question is.

The facts are interesting. A Dr Stephen Thaler (a frequent visitor to the courts for AI related claims) had asked the United States Copyright Office to register him as the owner of the copyright in a work of art which he said had been “autonomously created by a computer algorithm running on a machine” named the ‘Creativity Machine, which he owned.

He was, in other words, running a line of argument which said that he should be regarded as the owner of the work not because he had done anything creative in relation to it, but because he had engaged a machine to engage in creativity on his behalf. In the same way that one might engage a creative agency to work up artwork for a brand or product.

The USCO, and now the District Court of Washington DC, declined that line of reasoning. In their view it cannot succeed because, in the USA, they consider that US copyright “protects only works of human creation” and that “human involvement in, and ultimate creative control over, the work at issue was key“.

Or, put simply, the court came to a conclusion that copyright cannot exist in works created by a machine without some degree of ‘human involvement’ and ‘creative control’. So works generated solely by AI do not attract protection.

That’s big news for any industry / organisation which had plans to replace human creatives with generative artificial intelligence of one kind or another, because it means that there is a strong possibility that (in the USA at least) the works of that artificial intelligence would fail to attract copyright protection. Which, in practical terms, means that it would be open to copying, reproduction, redistribution, and other forms of infringement by third parties with no possibility of redress.

However, I think that it’s important to read this judgement with a small note of caution, it’s important to recognise that Dr Thaler was, to use a technical term, ‘going large’ with his application to the USCO and was trying to record the Creativity Engine as the literal author of the work in question. That’s a far cry from applying for protection of a work where AI has been used to assist and streamline the work of a human artist, and it isn’t a conclusion that precludes judges in other cases from concluding that the act of ‘prompting’ a generative AI for particular outcomes is an act sufficient for that person to be said to have ‘creative control’.

But your key takeaway from Dr Thaler’s latest expedition to the courts is that the direction of travel in the USA – which may well act as a significant nudge to the rest of the western world – is one of reluctance to grant copyright protection to images and works of art created by generative AI. Which as significant implications for your IP protection strategy if you intend to rely on that kind of asset in your productions.