Can work produced by machine learning software create issues under copyright laws? Can copyright laws be extended to computer-generated work as well? Benjamin Sobel, a law and technology researcher from Harvard, has discussed this quandary in a research paper titled Artificial Intelligence’s Fair Use Crisis. He has said that by ingesting data from multiple resources, AI programs can now create music, write novels and even edit and make movies. That is how Deep Mind’s Alpha Go won over reigning (human) champion Lee Sedol — by ingesting data from 1,60,000 amateur games. AI programs learn from a vast body of data generated by humans to come up with the optimal result. Then again, as Sobel notes in his paper, machine learning is an artificial intelligence technology with immense potential and a vast appetite for copyrighted works.
However, can a machine be awarded intellectual property rights? Given the pace at which AI systems are generating unpredictable outputs by ingesting vast amount of data, it is difficult to attribute who is the creator of the final work. Past rulings governing IP laws in the US and Germany deem that only works created by humans come under the purview of copyright laws. Does this indicate that works created by AI fall outside copyright laws because they are not created by humans? Given that AI systems would be governed by multiple players, determining ownership of AI-produced content is tricky.
According to the guidelines set by the US Copyright Office, ‘non-humans’ are not natural persons and cannot be held legally responsible in a court of law. Hence, they can’t be considered ‘authors’.
Can Work Produced By AI Be Assigned Authorship?
This paper, authored by K Hristov questions IP laws in the field of AI. According to Hristov, when it comes to assigning ownership to AI-generated works, one can consider three main players:
- AI programmers
- Owners, that is, large companies and financial investors in the AI sector
- end users
The author also brings another perspective into view — when awarding ownership of AI-generated content, how would society benefit most? If it is awarded to the programmer or the end user or the investor?
To better gauge the societal impact of each party, we must first determine the ultimate goal of assigning copyright of AI-generated works to human authors. Another important point is that since non-humans can’t be held responsible in court, ownership should be assigned to humans. Therefore, providing ownership rights to AI developers or end users can be useful for the growth of the sector. Hristov further observes that financial incentives should be awarded to those who make the greatest contribution to the development of AI. This would thus foster innovation and spark a renewed interest in research and development of AI.
Can Legal Standing Affect The Future Development Of AI?
Indian IP attorney Anmol Khurana of Khurana and Khurana law firm noted in an article that there are only two ways in which copyright can be bestowed to AI — first, it can be completely ruled out since there is no human involvement. Or second, it can be attributed to the developer. Going forward, copyright laws will have a bearing on the development of AI. Khurana observes that with the usage of AI increasing daily, machines will become better at artistic work and the lines between human and machine-generated content will blur. Dr Paul Lambert, adjunct lecturer and author, observes that as technology becomes increasingly commercialised, AI ownership should be governed either on a case-to-case basis or grouped by category.
- Even the data that machines ingest is subject to IP rules.
- Copyright issues would play a huge role in governing robots and drones – that operate semi-autonomously, especially in areas of surveillance where drones collect pictures. Would these activities carried out by autonomous agents come under the purview of copyright laws?
- Drones are being widely deployed in law enforcement, domestic services such as cargo delivery, emergency operations and humanitarian operations and are subject to regulation and licensing. They are directly controlled by an operator but gradually there is a rise in autonomous flying and sub-aquatic drones which are also used in surveillance for collecting data. These drones do not have a human — in the loop which begs the question, would the activities and decisions taken by these drones have any legal bearing.
- Given the vast amount of resources being plowed into the development of drones, robots and AI research and commercialisation, it is time enterprises devote energy to understand whether these activities would be protected under copyright laws.
Going forward, there needs to be stricter regulations regarding machine-generated work. As robots gear up to transform the economy and the usage of drones increases, enterprises and governments should tackle the issues related to copyright laws and settle ownership of AI-generated content. On the other hand, AI is wending its way into legal departments and is affecting the future of law. India is seeing a startup boom in legal tech with vertical AI startups such as Casemine and Pensieve operating in this area. Perhaps, the issue of IP and copyright laws will be addressed by legal tech companies working on NLP and AI.
Try deep learning using MATLAB