AI is cleary having a major impact, but not every situation is novel
Stop treating all issues involving AI as ‘unprecedented’, advises Ross Nicol
In a previous article for The Scotsman I highlighted how Artificial Intelligence (AI) is changing the way the legal profession delivers services. It is equally true that AI is impacting the legal agreements which clients are entering into.
We have seen AI considerations move rapidly into the mainstream of corporate transactions and commercial contracting, with parties increasingly seeking to understand and address how AI is being used, what risks that use creates and how those risks should be allocated.
AI is now a feature (and increasingly a focus) of due diligence exercises. A buyer of a target company, which makes use of AI tools as part of its operations, will want to understand which AI solutions are being used, the associated terms and conditions, how those solutions are trained, how they are used within the business (or as part of outward-looking service delivery), and the AI policies governing use. There may also be diligence questions regarding output incorporating AI-generated material – a buyer will want to understand how the business satisfies itself that it has the necessary rights to use that output.
Where AI is fundamental to the target’s product or business model, the diligence exercise can become considerably more extensive. A buyer may want to understand things such as the provenance of training data; the models and third-party AI services incorporated into the solution; contractual dependencies on AI providers and compliance with applicable AI regulation. There are material intellectual property, data privacy and confidential information considerations. The importance of such diligence is only likely to increase as the legal landscape catches up with developments in AI.
The diligence findings often flow into the transaction documents, with buyers seeking specific warranties, indemnities or other contractual protection concerning AI.
We are seeing a similar development in commercial contracts. Customers increasingly propose dedicated AI clauses or schedules, particularly when procuring technology services. Larger organisations will increasingly have a standard set of AI clauses which seek to address key AI risks.
We have observed that often these terms, driven by caution around AI and the lack of a clear “market standard”, can be onerous (and even impractical) for the supplier, for example stifling their legitimate need to continue developing and improving their solution.
Having been immersed in advising on AI for the last few years it’s clear that many clients rightly remain cautious regarding AI and the potential impact on a transaction. A key part of our role is to help clients understand the significance of AI on the target business and the transaction and to help them identify and navigate genuine issues of concern.
A further observation is that whilst AI creates genuinely new legal and commercial considerations, not every issue involving AI is novel. Before generative AI became ubiquitous, for example, Software as a Service negotiations regularly addressed whether suppliers could use customer data, usage information or outputs to improve their services and products.
The same is true of many other AI questions. Intellectual property, confidentiality, data protection, cybersecurity, subcontracting, regulatory compliance and liability have long been central to technology transactions. AI may alter the analysis, and in some cases materially increase the risk, but it does not make those established principles irrelevant.
The challenge for businesses is therefore to strike the right balance, and parties should resist treating every issue involving AI as unprecedented simply because it carries an “AI” label.
Ross Nicol is a Partner, Addleshaw Goddard