Artificial Intelligence Law
How do you protect your AI?
Legal protection for AI is somewhat limited, and more so in Australia than other countries. However, there are still avenues for implementing contractual rights even where regulation breaks down.
Copyright. While a number of countries have introduced provisions into their copyright acts (including New Zealand), Australia is not one of them. Even those countries who have implemented AI oriented provisions into their copyright acts have only put in place rather rudimentary protections. The related body of law is limited. The algorithms developed to build AI are protected by copyright but the ownership of outputs, such as neural networks, is lacking. Keep in mind, the original author of computer code, including AI algorithms, is likely to be the owner. In which case the code must be assigned in writing to be effective. If there is any uncertainty around your ownership of the copyright code, call us to discuss.
Protecting the outputs. Protecting your AI requires a little more imagination given the limited copyright protection. AI developers will likely want to include provisions in their terms of use to limit the purpose for which the output from AI can be used. Notably, in many cases limits should be placed on the use of APIs, through API agreements, and prohibitions on reverse engineering should be included in any terms of use.
Of course, there are a range of technical protection mechanisms which you can, and should, also put in place to protect your business.
What are the key commercial considerations in Artificial Intelligence Law?
Contractual clauses should be tailored for AI transactions.
Most of the talk these days relates to the limitation of copyright law, and in particular, the neural networks developed by AI algorithms. However, there is more to Artificial Intelligence Law than just copyright. The following represent some of the common considerations for those implementing or developing AI capabilities and products.
Data rights and privacy. Critical to training or using artificial intelligence are data rights. Make sure you have provisions in your contracts granting you the right to use data collected, or provided to you, by your clients. When considering privacy, note that personal information includes information about an individual who is reasonably identifiable. With improvements in data processing, information which was once de-identified may no longer maintain that status, particularly when combined with large data sets.
Indemnities against bias. There have been some spectacular (or perhaps notorious) failures of AI. Bias is easily input, and that bias can result in risks to both reputation and profit. From the Artificial Intelligence Law perspective, you will want to be sure liability associated with that bias is limited and appropriate indemnities are obtained.
Liability, disclaimers and limits on use. AI is not perfect, nor are the humans which feed it. For that reason, it is important to ensure appropriate limits on liability have been agreed. In some circumstances, particularly when working with MVPs, the purpose for which the software can be used may be limited to non-commercial use, or for decision support not decision making. In this way Artificial Intelligence Law is like broader software development law.
What are the key human rights and regulatory considerations in Artificial Intelligence Law?
Reputation risk and regulation are key considerations when implementing AI capabilities.
Emerging regulation. Scholars and policy makers throughout the world are continually debating the role of regulation in Artificial Intelligence Law. At the moment it’s a lot of talk and little action by governments. However, no doubt that is close to changing. Notwithstanding the limited regulation on AI there are still broader compliance issues, such as privacy and human rights, which need to be considered. Furthermore, developers will need to consider the reputational risks associated with AI use cases which do not meet the general public’s moral or ethical expectations. Governance issues such as these will continue to arise and broaden the scope of Artificial Intelligence Law.
Human rights, discrimination and exploitive practices. The Human Rights Commission also points out concerns over exploitative and discriminatory marketing practices which may be automated and narrowly focused. There are concerns about intrusive surveillance in the workplace and we are already seeing problems arising from automated debt collection. In financial circles we have had, and are conscious of, discrimination in the provision of financial products and the assessment of credit worthiness. Not only will these considerations likely be the targets of regulation (with calls for the establishment of an AI Safety Commissioner) but also represent a significant risk to reputation if carelessly implemented.
Where can we find an IT Lawyer?
Where else would an IT Lawyer base themselves other than in the trenches with other tech entrepreneurs and start-ups.
You can find us at River City Labs in Fortitude Valley, Brisbane. If you have never been to River City Labs it cohabits with the Queensland Chief Entrepreneur and proclaims to house over 150 tech businesses. Furthermore, it is one of the most active venues for venture capital in Brisbane and holds regular pitch nights.
