Patent law then, now, and in the future (with AI)
Patent law has evolved over centuries, adapting to a world in constant change. Now it faces the next major technological development – AI. But what role can AI actually play in patent law?
From Venice in 1474 to today’s patent system
In antiquity, ideas began to emerge that technical knowledge and creativity were in the public interest and worthy of protection. The fall of the Western Roman Empire was followed by a general decline. However, in the late Middle Ages, knowledge and technology came to be valued once more as an urban middle class emerged. In the republics of Venice and Florence, inventors were granted temporary monopolies. The same was true in England, though in that case by a monarch. This gradually became more common.
The first general patent legislation was enacted in Renaissance Venice in 1474. The purpose and main features of the law are familiar: to stimulate innovation and promote the public good, inventors could obtain a time-limited exclusive right in exchange for publication; in the event of infringement, they could count on compensation and the destruction of the copies.
Although patents essentially constituted granted monopolies and privileges, patent law survived both the British Parliament’s curtailment of royal power and the American and French revolutions’ quest for freedom. By the early 19th century, many countries in Europe had adopted their own patent legislation, and following its unification, Germany established a successful patent office in the 1870s that carried out preliminary examinations of patentability.
However, the laws of different countries differed, and discrimination against foreign companies created barriers to cross-border trade, a key driver of prosperity. Added to these challenges were the extra bureaucracy and high costs resulting from the territorial nature of patent law and national boundaries. The legitimacy and effectiveness of patents were called into question. Many long and difficult negotiations took place between countries with a view to internationalising and streamlining patent regulations. In stages, the export industries obtained what they had, with good reason, been demanding.
The Paris Convention (1883) introduced national equal treatment and the right of priority (the right to apply for a patent for the same invention in other contracting states within one year without destroying novelty). After two world wars, patent law was harmonised, streamlined and further extended through the European Patent Convention, with a single examination procedure across Europe (1977); the TRIPS Agreement, setting out minimum standards for IP protection in global trade (1994); and the UPP/UPC, providing for unitary patent protection and a common patent court within the EU (2023).
In recent years, patent law has been called into question in climate negotiations and in the field of public health. However, the reality that the major resources lie within the private sector (rather than the public sector), and the argument that the high-risk development of green technology and new medicines requires financial incentives, have generally carried the most weight. Patent law, which offers such an incentive, remains in place.
Is AI patent law’s next challenger?
The next challenger is artificial intelligence, or AI. Will it become the nemesis of patent law? It does not appear so. For Europe, the EPO’s Boards of Appeal have ruled that only persons (with legal capacity), and not machines, can be inventors in the legal sense. The question is whether AI might instead replace us in the application of patent law?
AI as a tool for legal professionals
Given a question and a large body of material, AI can quickly search for and find, sort, summarise, compare, question, suggest or create. It can turn over forgotten stones or simply provide inspiration. Only the imagination seems to set limits on what AI can be trained to do.
Large language models, or LLMs (which also form the core of chatbots), can be excellent tools for rapidly improving the quality of language when legal texts such as contracts, applications, submissions or decisions need to be drafted or translated. At the same time, their statistical models and predictions clash conceptually with the nature of the application of the law. Even if the prompt is good.
AI can provide support – but it cannot take
responsibility
Patent law is a legal discipline (in a technical context). Whilst the rules and principles of law strive for universality, the application of the law must be specific. Rules must be interpreted, evidence evaluated, considerations taken, and interests weighed against one another. The answer to a legal question should be reasonable, not merely probable. Ultimately, the answer is determined by a court. Those who commission or propose a legal text should realise that and take it into account.
People with expertise and sound judgement are needed to apply the law. Also to understand, evaluate and appropriately use AI in those processes (or, in the first place, to train the machines). The support that AI provides needs to be assessed at some stage, or else it will end up as guesswork. For the client, there should also be someone available who can take genuine responsibility should things go wrong.
The wave of AI sweeping across the world is exciting and full of opportunities. At the same time, it raises questions that demand answers. Pressing questions about data protection and confidentiality, and existential ones about making humanity dumber and destroying the climate. For centuries, the order and transparency provided by patent law have stabilised and accelerated technological development. Machines, such as AI, can help to trim the sails of patent law. Their role is not to replace the captain.

Claes Almberg
Head of Legal Attorney at Law UPC Representative