The Federal Court of Justice, Bundesgerichtshof decided in a ruling issued on June 11, 2024 (AZ X ZB 5/22) that Artificial Intelligence cannot be recognized as an inventor. Only a human can file for an AI generated invention. The DABUS cases are pro-corporate attempts brought by the Artificial Inventor Project seeking intellectual property rights for AI-generated output “in the absence of” a traditional human inventor, but the courts are not buying it, and the result is and will always be the same, you need a human name on a patent, regardless of how little input the human made in generating the invention.
Normally, to register a valid patent for an invention, you need to prove “substantial human contribution”, so even human inventors who are for hire would need to have their names on the patent. Previously, German courts were split on the issue. Now, the Bundesgerichtshof has resolved the split by removing the requirement for “substantial contribution” by a human.
What the Bundesgerichtshof is doing basically is to tell the courts to stop obsessing over the degree of contribution of human versus machine input. It is unnecessary to examine how much of the process of invention has been automated. Everyone agrees that machines cannot invent anything coherent entirely on their own and if they could, it would take a human to decide whether something was invented, so without the human, there is no invention. It matters very little what technology you use to come to the conclusion that something is an invention.
So humans will continue having their names on the patent, but they won’t need to prove they never used AI to generate parts or the whole of the invention. The requirement for a human inventor is simple, if someone uses the patent without permission, a robot cannot file a lawsuit and you can’t sue a robot for infringing on your IP. A robot can’t assign rights to anyone because it is a corporate asset. Assets are owned, they have no agency or the capacity to consent. Given that corporations are not recognized as inventors, they need at least one precedent where a corporate asset can replace the actual human inventors. There is no other goal in trying so desperately to remove the human inventor requirement. Right now, if you are for hire, you already consented to be deprived of your rights with or without AI. Even before the advent of AI, it was customary for the CEO of a corp to put their own name on the patent even though 9 other employees made the invention and all their names are not necessarily on the patent. If AI becomes an exception for the human inventor requirement, it will be another step into corporate appropriation of human work.
Luckily, DABUS is an extremely weak case, well publicized all over the media but very weak, from here, I’d say the DABUS claims begin to border on frivolous at this point. How many times in how many jurisdictions can a plaintiff lose the same case, before being declared a vexatious litigant? The fact that there seems to be unlimited money to bring an unlimited number of the same version of unsuccessful case is telling. I think the courts have better things to do right now.
It is another way to say that human users will always own the rights to AI generated output, so long they have provided the most minimal of input in a prompt and made a final call regarding the generated output. If no human was involved in the generating of an invention, then it wouldn’t be possible to register it. AI platforms are simple tools, no different from other applications you may have used to create your IP. Basically, the German courts are instructed to stop caring what tools and mediums an inventor used to create the IP, be it Microsoft Word, a gas-stove, a shovel, Ableton Live, a tractor, a fork, artificial intelligence, an X Box Kinect, a hair-brush, or any other tangible or intangible object.
Most often, tools may have not been used at all for a valid invention, humans often have an instant vision of something they need to use at a specific moment, but that doesn’t yet exist. You imagine it, you make it, you use it and if you want to make money with it, then you patent it. Otherwise, I believe the majority of existing inventions are not even patented. Conversely, the majority of patented inventions are so abstract that they may be as good as useless. Inventions come from a specific need. Patenting whatever is patentable and isolating molecules from efficient systems (i.e. things from nature) has proven time and again to be counterproductive old-world mentality, but this is a subject for another post.
What is true for inventions is even more true for music or script-writing for example (I hope coding as well, because I will need to code soon and will not hesitate to use AI.) Humans hear music in their heads and our minds create multiple scenarios faster than the speed of light. Of course we need tools to organize all this information and take it out of our heads in coherent form from time to time. AI is here to facilitate and accelerate human creation and productivity. Corporations don’t seem to like this. Before the advent of AI nobody cared whether you compose on a MacBookPro, on a phone, with a pen and a harmonica, or by recording your washer and dryer to make beats. Bottom line is, we have all these billions of machines and tools, but it takes a human to make shit and mainly to decide if it has been made at all.
In Australia, a court opened the door for DABUS being registered as a non-human inventor. I haven’t read the decision as it is still appealable, so the law is not yet settled in Australia, but when it will be it is unlikely to be any different than the status quo https://artificialinventor.com/wp-content/uploads/2021/08/Thaler-v-Commissioner-of-Patents-2021-FCA-879.pdf
On a first glance, it seems that the Australia courts keep requiring the need for a human to own and control the invention, but in the decision there is a discussion as to whether AI can be named an inventor for the sake of being named an inventor, even though only a human can file a valid patent and be a patentee, regardless of the number of “inventive steps” the machine has taken, or any thought processes a human has had. As I explained above, nobody cares how a human applicant got to the invention. Practically, it is the court thinking out loud philosophically while nothing really changes. To cite paragraph 12 from the judgment
To sum it up, the Australian court says AI could be a “sole” inventor, but you still need a human to take credit for the AI’s work in order to register the invention and derive any economic benefit. After all these mental acrobatics, it looks like we are at the exact same place we started out.
It’s all great stuff, and it all points to the same place. Humans will own everything AI generates. When you see a court discussing definitions in the dictionary, it means the law is no longer of any help and everyone is completely lost. Here we have one of those moments.
I didn’t know that “computer” initially referred to a human who made computations. So, a human can be a computer, but a computer cannot be a human. Got it. What helpful information to start the day.
