There is no question that the proliferation of artificial intelligence (AI) and machine-learning is fundamentally transforming how many industries operate. From insurance to healthcare and manufacturing, embracing the potential of AI is no longer a competitive advantage, but rather a business necessity for survival in an increasingly competitive global marketplace.
Yet, there are considerable commercial challenges of actually using AI in practice, especially in the interaction between how and when users of AI-based solutions and their AI suppliers interact. The vast majority of standard software agreements simply don’t cover many of the often complex areas
AI reaches.
“The big difference is that with software you can explain to a developer what you want to do and they will create a piece of software to achieve that, which can be clearly identified, but with AI that isn’t the case. Fully understanding what constitutes an AI solution and distinguishing it from the customer’s own data is a much more complicated process,” says Tom Lingard, partner and head of intellectual property (IP) at Stevens & Bolton, a full-service UK law firm.
Enterprises are beginning to realise that AI contracts require a different approach to traditional software agreements
Unlike traditional software, it is difficult to delineate clearly which aspects of the trained AI are brought by the developer and which are specific to the customer, as well as the extent to which those should be owned by either party. It is not easy to separate AI collaborations and interactions, leading to potential IP liability and ownership problems.
A multitude of IP issues are set to arise as AI continues to be instrumental in creating innovative systems and it becomes extremely difficult to discover what is being provided by each party to the transaction. If an AI platform or solution is being created as part of a collaboration then there is a question around to what extent
the person or company providing the data should share ownership of the finished platform.
Once the AI system is up and running, a wide-ranging conversation around which party is providing the data and training the AI needs to happen, alongside to what extent data and analytics can be extracted or taken away if the collaboration comes to an end. Only a few industries won’t need to plan for the legal risks associated with IP and AI.
“In theory, AI platforms can be helpful to any type of industry. There are obviously some where it is going to be inherently more important, especially those operating in highly technical areas where AI is being used to analyse large amounts of data to support or generate patentable inventions,” explains Mr Lingard.
There is also a risk in terms of where the data for training AI is coming from. Some AI solutions may make use of data mining to uncover relevant data from third-party sources, which may not have given permission for that work to be copied.
As the AI system itself can’t incur liability, the parties involved in its creation and use must be clear about who would be liable for IP infringements as they surface. In this complex IP environment, Stevens & Bolton can help firms develop a comprehensive IP strategy for AI platforms, which helps prevent damaging IP breaches and protects innovative AI solutions as they are built.
“Some clients will be in the vanguard of AI projects with others being approached by firms offering AI platforms to improve their day-to-day business. We understand where clients are on this journey and equip them with the tools to recognise how these risks can apply to their business,” says Mr Lingard.
Businesses will need advice on everything from how correctly to ascertain IP ownership to defining liability conditions and discerning which party is responsible for AI actions that are not 100 per cent controlled by direct human input.
“Enterprises are beginning to realise that AI contracts require a different approach to traditional software agreements. AI has the potential to be very useful for businesses, but its unique IP challenges need to be addressed or costly IP issues may arise,” Mr Lingard concludes.
For more information please visit www.stevens-bolton.com
