Challenges posed by Artificial Intelligence (AI) and Generative AI in M&A Transactions need to be addressed with an interdisciplinary team. This Insight newsletter addresses some key issues dealmakers must keep in mind.
Since the launch of Dall-E and ChatGPT last year, Artificial Intelligence (AI) in general and Generative AI (GAI) in particular are omnipresent. The breakthrough of AI technology is not a product of chance but based on vast progress in availability of affordable computing resources, big data and open source AI models. Switzerland, with its leading universities and fast growing startup and tech scene, offers fertile ground for AI research and startups.
From an M&A perspective, AI is becoming more and more important. It is not just that AI has started reshaping due diligence processes – going forward, the acquisition of, and entry into joint ventures with, businesses active in the AI and GAI space will be an integral part of day-to-day M&A. With AI increasingly becoming a cornerstone of digitization overall, AI will be integrated into many businesses outside the tech and start-up space – and should thus be in focus for investors in respective transactions.
What to look for when looking at AI/GAI
AI is a novel technology that poses particular challenges that must be looked at comprehensively from different perspectives:
- From an intellectual property (IP)
perspective, it is important that the company owns the rights to
the model and has secured that neither the model nor its output or
the training data infringes third party rights. With the datasets
and the models' complexity growing further, this task becomes
more and more challenging. Given that the legal protection
available for databases is limited and the fact that many AI models
heavily rely on open source components, it is key to conduct
appropriate due diligence and implement appropriate protection in
the transaction agreement.
- Another focus area is data protection, in
particular in Europe and Switzerland. The risks in relation to data
protection may be quite different for an AI solution that analyses
health data for recognizing brain cancer than for an AI solution
that helps finding defects in car parts. Thus, a risk-based
analysis is always key when allocating resources in any given
transaction. In particular in sensitive industries, which largely
correspond to those with specific professional secrecies (e.g.,
banking, health, etc.), data privacy should always be a key focus
of any transaction.
- Similarly, cyber security concerns are
increasingly in the spotlight, considering the countless attacks
both from government sponsored cyber warriors as well as private
actors. The attacks may occur both on the level of the
infrastructure or the AI model and its user interface itself and
related risks must be appropriately addressed in any AI-related
- The increased scrutiny from regulators around the globe should also be considered when investing into AI-based business models. The European Union's AI Act is currently in its final stage of legislation and is set to become the first comprehensive and cross-sectoral regulation of AI. Already now, businesses should prepare for the AI Act, as its key principles are largely known. In addition, specific sectoral requirements may apply. As a first mover, the Monetary Authority of Singapore (MAS) already mid-2022 released its "Principles to Promote Fairness, Ethics, Accountability and Transparency" addressed at financial intermediaries using AI tools. Thus, potential investors should assess now whether AI tools meet the increasing demands from regulators (e.g., as regards explainability, transparency and non-discrimination).
While there is no AI-specific regulation in Switzerland, Swiss regulators are increasingly looking into AI. For example, FINMA has recently created its own competency center for AI and has also conducted a survey among all authorized financial institutions on their current use of AI, in order to get the necessary data basis to start its regulatory work. In addition, in the current revision of the Copyright Act, the Federal Institute of Intellectual Property is considering whether the bill should be supplemented with an obligation to remunerate the original author for the use of journalistic content through AI applications.
Challenges posed by AI/GAI are best addressed with an interdisciplinary team
Addressing these challenges in an M&A transaction – identifying the client's needs and risk appetite, conducting a focused and effective due diligence and negotiating the transaction agreements to address the identified risks with adequate reps and warranties as well as indemnities and covenants – requires an interdisciplinary team of M&A, IP and tech as well as regulatory experts
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.