Executive summary
A decision to adopt AI can raise fundamental and moral issues for society. These are complex and vital issues that are not typically the domain of lawyers.
We use cookies and other similar technology to collect data about you to allow us to deliver our online services, measure our website audience and improve your browsing experience. Full details on the cookies we use are set out in our Cookies policy. Please click OK to signify your consent to our use of cookies.
You can withdraw your consent by clicking “manage cookies” and following the instructions shown.
Part of our Artificial Intelligence briefing
Global | Publication | July 2018
Recent technological breakthroughs and digitalisation have led commentators to describe this time as ‘the fourth industrial revolution’ and the ‘robot apocalypse’. Artificial intelligence (AI) and machine learning are significant advancements in this revolution.
As industry players scramble to keep up with these developments and seek to implement their own AI innovation, intellectual property (IP) rights have become a prominent area for speculation and scrutiny.
AI has the potential to be both the creator and product of innovation. IP law has thus far developed at a far slower pace than technology, which creates a number of legal challenges for innovators and investors alike. This article will identify and consider issues that AI will create for the IP rights of those operating in Australia.
Copyright protects literary, dramatic, musical and artistic works, and has its roots in creative industries as opposed to technology. This creates challenges when using copyright to protect the output of AI; the drafters of copyright legislation had no way of knowing that the near future would involve autonomous vehicles and robots with human capability. You may recall the infamous ‘monkey selfie copyright dispute’, where a number of selfies taken by inquisitive macaques was the subject of heated debate and litigation as to ownership. This case foreshadowed the primary issue for AI and copyright – where there is no human author, who owns copyright in a work?
The legal position in Australia is that copyright requires human authorship. This was considered in the ‘white and yellow pages’ decision, where it was held that copyright did not subsist in telephone directories that were largely computer generated. This is problematic when considering that AI is already being utilised to create musical and artistic works, and produce images. Recently, a facial-recognition algorithm that scanned data from the Dutch painter Rembrandt was able to generate a 3-D printed painting in his style.
The US has a similar approach. Shortly after the ‘monkey selfie’ copyright dispute began, the US Copyright Office issued a public policy stating ‘the office will register only works that were created by human beings’. The US District Court agreed with the policy, resulting in a finding that no copyright exists in the works. Of course, it might be debatable whether selfies can be classified as artistic works in any event, but that’s for another article!
The approach being taken in Australia and the US is not, however, being followed universally around the world. This creates confusion for those operating globally when considering other countries’ approach to copyright.
The Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), an agreement between member countries of the World Trade Organization regulating the protection and enforcement of IP rights internationally, states that ‘copyright protection shall extend to expressions and not to ideas, procedures, methods of operation or mathematical concepts as such’, but does not descend into detail regarding human authorship (or lack thereof). This creates room for national legislators to form their own policy and take differing approaches to a fundamental copyright principle. For example, UK copyright law states that where works are computer-generated, the copyright of the works is held by those who made the arrangements necessary for their creation. New Zealand’s Copyright Act also suggests that original works created by software or AI are protectable. This would lead to a better outcome for those seeking to establish copyright in their AI-generated IP rights than in Australia.
However, it does not seem essential for Australia to have an updated IP ecosystem that encourages investment in new technology, as other global superpowers with similar systems, like the US, still attract investment and innovation. It remains to be seen whether Australia will change its policy in light of others’ treatment of IP rights. Although other countries have taken a more reactive approach to AI, this does not necessarily mean it will inevitably be the case in Australia as global superpowers like the US have continued with their current approach.
Patents provide a time-limited monopoly right for a new invention. In order to qualify for patent protection, the invention must be novel and involve an inventive step when compared with the prior art base as it existed before the priority date of the patent application. Novelty and innovation are assessed on a worldwide basis, meaning patent filings must be co-ordinated internationally to avoid the risk of losing patent rights in other jurisdictions.
Australia may be a more advantageous jurisdiction to seek patent protection for those concerned about their innovation meeting the inventive step threshold, as it offers an ‘innovation patent’. This unique alternative to the standard patent is intended for technology that does not meet the standard inventive step threshold. Instead, the technology need only be an incremental improvement to existing technology, that contributes to the working of that technology. Consequently, it offers a lesser protection of 8 years (as opposed to the standard 20). This may be an ideal alternative for those seeking IP protection for the outputs of their AI if there are concerns that their invention is not a breakthrough in technology or might otherwise fall short of the standard patent threshold, but are valuable nonetheless.
The future of innovation patents was placed in question recently when draft amendments to Australian IP legislation was introduced including provisions that sought to transition innovation patents out of Australia’s patent system. However, the Bill introduced to Parliament omitted this section, meaning innovation patents are safe – for now.
The subject matter of the invention must also be ‘a manner of manufacture’, which is often a further obstacle for the output of AI to overcome. As a starting point, the manner of manufacture must involve a physical effect; this does not have to be a tangible result, and may exist in the application of a physical device such as a computer. Consequently, a mathematical algorithm (which is often what AI technology essentially involves) by itself is not patentable.
It has been a matter of contention internationally whether computerised processes are patentable. This was considered by the Full Federal Court of Australia in the ‘RPL Central’ decision. The Court implied that the output of AI may be patentable, but drew a distinction between using a computer as an intermediary to merely perform a method (which is not patentable), and there being an invention in the way in which the computer carries out the scheme or method (which has the potential to be patented).
This provides hope for the use of patents to protect the products of AI technology in Australia. It does, however, create uncertainty as to which inventions will actually be patentable. A computer process that assembled text in Chinese characters using a non-Chinese keyboard was held to be patentable, but an automated safety assessment was not.
AI has the capability to perform a wide array of functions, from the mundane to the ground breaking. The threshold of inventiveness required for different outputs of AI to qualify for patent protection is unknown. Under the current test, however, it is clear that those considering patent protection as part of their AI IP strategy in Australia must ensure the innovation lies in the use of the computer itself as opposed to the idea or business method.
Registration is not necessary under Australian law for IP to be protectable as confidential information, which is generally protectable to the extent that the information has not entered the public domain. This may be a helpful fall back method of protection for AI technology for which traditional IP protection is not available, such as pure data or trade secrets. However, confidentiality is not codified and only arises from the existence of a contractual relationship or general obligations in equity.
Many forms of AI technology operate more effectively when information is freely available. Open-source data sets have positive impacts upon the scope of solutions that AI technology is able to generate and the quality of the machine’s analysis. This creates tension, however, when considering the obvious risks it creates to preserving what is often now an organisation’s most valuable asset – its confidential data. The settlement reached between Uber and Waymo in their trade secret battle was reported to have a mammoth $314 million value.
Although Tesla famously sacrificed its IP rights for electric vehicles in the interest of progressing others’ electric car programs by implementing an open innovation strategy, most organisations choose to take a more conservative approach to sharing their IP. If traditional IP rights are unavailable, organisations must rely on non-disclosure obligations to protect their AI innovations. Once information is in the public domain, an interlocutory injunction may be obtained to restrain its use; however, this carries obvious time and cost consequences. In the absence of a statutory regime, it is essential that any contractual arrangements involving the use of confidential information deal thoroughly with the event of a breach to ensure the owner has robust rights and provide a framework that sets clear parameters as to how such information may be used.
The effective protection of confidential information is not just a question of having the right paperwork, but involves creating a culture within an organisation where there is ongoing IP education. Processes must be put in place and applied to ensure IP rights are identified, protected and maintained, as once a breach occurs, confidentiality becomes increasingly difficult and costly to restore.
Organisations must not only be aware of AI’s utilisation, but the uncertainties it creates for their IP rights in all corners of the globe. Innovators cannot afford to be idle in their protection of AI technology and must vigilantly develop a strategy to ensure their rights are adequately protected.
As the application of existing IP regimes to the output of AI at this point is undeveloped and often ambiguous, innovators should seek to form layers of IP protection using copyright, patents, and confidentiality to build an arsenal of IP rights. Although Australian law is subject to change, the Productivity Commission report into Intellectual Property Arrangements recommended that Australia work with like-minded countries to achieve more balanced IP arrangements that facilitate the further production of creative works and innovation, which may be an indication of the direction that Australia’s IP system is to take.
A decision to adopt AI can raise fundamental and moral issues for society. These are complex and vital issues that are not typically the domain of lawyers.
AI is a field of computer science that includes machine learning, natural language processing, speech processing, expert systems, robotics, and machine vision.
AI will need to meet certain minimum ethical standards to achieve sufficient end user uptake, varying according to the type of AI and sector of deployment.
Courts in a number of countries have already had to address a range of legal questions in relation to the automatic nature of machines and systems.
The key question businesses need to consider is whether deploying AI will result in a shift of ethical and legal responsibility within their supply chain