Last August 18, 2023, a ruling of US District Court for the District of Columbia in the Civil Action No. 22-1564 (BAH) denied copyright protection to an image generated with an AI in execution of the input (‘prompt’) of the user. At a first glance the ruling looks correct, however it is questionable because there are are many examples of copyrighted non-AI- generated content made without human intervention. The main problem with this ruling, indeed, is that it looks at the matter from the (wrong) perspective of the ‘AI subjectivity’ rather than the economic value of the final product by Andrea Monti – initially published in Italian on Strategikon – an Italian Tech Blog
Before entering into the details of the case it is necessary to explain why, beyond the journalistic interest in the facts, one should care about a foreign ruling. Judgements issued abroad have no value within the Italian jurisdiction, however it is not uncommon to read a reference to non national courts ‘in the lines’ of Italian decisions.
On the other hand, it is known for a fact that court cases involving high tech happen in the US first and only after a signigicant delay in Italy. It would be, therefore, hypocritical to deny that one has not taken into account what another judge has done in his own nationa court.
We could go on and on for decades debating about how inappropriate this behaviour is, how the outcomes from different legal system are easily reconciled and how unacceptable is the idea of being subservient to the judicial power of another State. The fact, however, remains and it makes no sense to ignore it.
That said and coming to the point, the case analysed in this column involves the claim of a plaintiff that used an AI to generate an image and, subsequently, tried to register it as a copyrighted work in the US Copyright Registry of the US Copyright Office (USCO).
USCO denied the request on the basis that the image lacked human creativity, a necessary requirement to claim copyright protection.
Dissatisfied, the ‘author’ challenged the refusal asking the court to overturn the initial decision, but to no avail. The court, indeed, upheld the USCO position and confirmed that a work generated by a computer without human intervention is not entitled to copyright protection.
As said, at first glance the Court stated a correct legal principle. However, the ruling should be taken with a pinch of salt because it is based on unexpressed assumptions that are not entirely consistent with the reality of digital content creation.
In general terms, way before than legal, the main (but not not the sole and the only) cultural mistake is psychological: giving software the status of a ‘subject’ instead of a ‘tool’, thus confusing (legal) individuality and (operational) autonomy. Just look at the pictures that usually accompany columns about AI: invariably they feature androids or other half-mechanised human-like creatures suggesting, on subliminal level, the existence of some sort of ‘consciousness’ in the machine. One might wonder what would happen if instead we were shown floor-cleaning robots wandering around a house or CNC milling machines buliding furnitures or mechanical components.
Although implicitly, the US ruling suffers from this perceptual bias because by denying the status of copyrightable work to the image it implies —actually, it says it clearly— that the final outcome has been entirely ‘made by’ a computer and, therefore, without human contribution. In other words, the non-copyrightability of the image is based on the lack of human creativity and not on the economic value of the computer output.
It this principle were to be generally applied, then also pictures taken with ‘burst-mode’ should not be entitled to copyright. Less protection should be granted to those photos taken with remote controls manning cameras placed in dangerous or inaccessible places (such as the cage of a hammer-throwing competition). Finally, no copyright at all should be acknowledged to those stills extracted from a video (technically feasible with a decent quality at least since 2015) where there is no human contribution to the recording of each individual frame.
The same argument should apply to those images processed with photoretouching software. Starting from a picture, a computer program turns a photograph into a pencil sketch or a painting of various styles or, more mundanely, ‘decides’ the best crop altering the real field of view or the most effective combination of exposure, contrast, highlights and so on.
‘The song remains the same’ when it comes to sounds. It is enough to cite the autotune or, making a broader reference, modellers such as the Kemper Profiler. In both cases there is a software acting as a proxy between the vocals of a tone-deaf singer or the electrical signals generated by a pickup of a musical instrument to make it sound in tune (the former) or create the sound —let’s say— of Ritchie Blackmore’s Fender Strat coupled with a Vox AC30 (the latter).
And if one wants to talk about automated voice generation, the new achievements of the AI that starting from a few fragments can imitate the voice of an individual, it is challenging not to see the similarities with the venerable Vocaloid.
Videomaking —and, thus, cinema— share similar issues. It is now technically and financially sustainable to ‘rebuild’ and put on set actors who are passed away or are too old for certain roles. Abba exploited these achievements in their incredible concert where the artists on stage are their digital clones, and the actors’guilds protested against the will of motion pictures producers will to AI-process artists image without their consent.
The circle closes with the impact of artificial intelligence on writing and, in particular, on translations. There is not much to add on the subject since the considerations already made apply, with due differences.
So, to summarise and conclude the argument, when it comes to the automation of content production, the issue to be discussed is not who their author is, but whether they are susceptible to economic use by those who, in one way or another, have produced them.
Put in these terms, the photographer who extracts an epic-shot of a sports performance from a ‘casual’ shot, the musician who generates a hit riff using an AI platform, or the writer who ‘automatically outs a very personal feeling all have the right to economically exploit this product and thus to claim the right (right) to control its reproduction (copy): copyright, then.
A different matter is moral right, which is almost never spoken of because it is of little interest in a world in which creativity —an individual and unrepeatable act par excellence— has become a product —a serial object— boring and reproduced countless times. All that interests us about a ‘creative work’ is its saleability and not the fact that it is the fruit of inner labour or years of study and application. What counts, like any ‘over-the-counter’ object, is the finished product, to be put in the cart, consumed and thrown away as soon as a ‘better-made’ one is found. In other words, a content made with a high technological contribution does not deprive it of its aptitude for economic exploitation and thus its protectability from (re)use by third parties. Therefore, what is needed is to distinguish the attribution of the creative act from the economic exploitation of a content in any way generated. Said differently still: one does not need to be an ‘author’ to ‘make a living’ from content.
We can, legitimately, be disturbed by the loss of the artist’s centrality in the creation of a work and worry about the further dumbing down of the individual who, having given up using his body, is now doing the same with his mind. But from a strictly economic, market point of view, the American judge’s decision -and the complaints of the AI catastrophists- make little sense. The phenomenon has occurred, it cannot be stopped, and so we can only decide how we want to adapt to a new reality.
Then, on the sidelines, we should ask ourselves whether it is legitimate to allow such important changes in our way of life to be decided by purely industrial and financial logics such as those of Big Tech but this, as the narrator said at the end of Conan the Barbarian, is another story.