What happens when an unstoppable force meets an immovable object?
AI is the unstoppable force in our metaphor: it’s quickly becoming ubiquitous across all industries as its integration into business models continuously deepens. Representation and warranty insurance (RWI) is our unmoveable object: it’s a fixture of the M&A landscape that many M&A parties rely on to grease the wheels of their transactions and get deals done.
The collision is unfolding in real time. AI’s ever-greater market penetration means that AI issues in M&A are no longer limited to “tech” deals. More and more targets, regardless of their sector, present significant AI-related commercial and legal considerations. Given that AI presents a variety of hard to verify risks, this raises increasingly complex issues around RWI. A question that follows is (1) whether reductions in RWI coverage around AI are likely, whether by AI-related exclusions or otherwise, and (2) if this could progressively push AI risk allocation onto due diligence and traditional contractual mechanisms such as indemnities.
To begin to provide an answer, we review the competing forces at play and some of the practical takeaways for both buyers and sellers. For more Fasken M&A thought leadership, visit our Capital Markets and M&A insights hub and subscribe.
The Cause: AI Risks Present Diligence Challenges
The nucleus of the tension between AI and RWI is an understandable concern on the part of insurers: confidence that all material AI risks in an M&A transaction have been identified and are fully understood can be hard to come by. On the one hand, AI systems can be difficult to audit, unpredictable in their behaviour, and dependent on inputs that may be hard to verify after the fact. On the other hand, AI risk may sit in such varied places as the relevant training data, the model itself, an open-source component, or the way the target deploys AI in its services, products and/or operations.
The Effect: AI-Related RWI Coverage May Narrow
Different potential corresponding impacts on RWI policies and processes are foreseeable. As a base matter, heightened insurer scrutiny of AI-related issues is inevitable and well underway. Narrower policy coverage is also possible, depending on the transaction and target.
A policy exclusion could be limited to a known issue, such as a particular data set, model provider, customer claim, privacy concern, or regulatory investigation. However, if the diligence record is incomplete, RWI insurers may push for broader language, such as terms excluding losses arising from the development, deployment, use, governance, or performance of the target’s AI systems generally.
It’s also possible that AI’s impact on RWI won’t be limited to a deal-specific approach. Some insurers may move toward more standardized AI exclusions, e.g., “absolute” AI exclusions removing coverage for claims connected to AI, such as AI-related misrepresentations, regulatory investigations, violations of AI-related laws, or the design, development, deployment, or failure of AI systems. On the other hand, AI-specific M&A insurance products could emerge.
Practical Takeaways for M&A Parties
Several practical takeaways follow for sellers and buyers as they look to adjust to this evolving state of play and include the following.
Practical Considerations for Sellers
Sellers can mitigate the possibility of RWI insurers imposing AI-related exclusions by making AI-related risks more identifiable and understandable. This could include, for example and as applicable to the particular target, a thorough diligence record of (1) privacy compliance, (2) intellectual property ownership, (3) training data origin and usage rights, (4) model development, testing and deployment, and (5) internal AI governance policies and ongoing monitoring and human oversight.
Simply put, insurers should be more likely to consider AI-related coverage where sellers can show not only how AI is integrated into the target’s business, but also how it is controlled. For example, AI model performance is likely to receive particular attention. If the target can’t demonstrate clear testing procedures, performance benchmarks, and detailed specifications for AI integration, an insurer may be reluctant to cover claims tied to inaccurate outputs, model underperformance, or unexpected system behaviour.
Practical Considerations for Buyers
Buyers will have numerous similar issues to consider. One is that the breadth of an exclusion may depend on how AI is defined in the RWI policy. If the definition is overly broad, an exclusion could capture ordinary software, analytics, automation, or machine-learning features that are central to the target’s business. Ambiguous definitions can also create coverage disputes. Expert input may be essential. Buyers should review proposed AI exclusions carefully and push for precise language appropriately limited to the relevant risks.
Should RWI coverage around AI-related risks narrow, AI-related risk allocation may need to move to conventional contractual mechanisms in the transaction documents. For example, buyers may need to seek remediation before closing, AI-specific representations, targeted indemnities, related escrows, and/or separate AI-specific insurance, if available. These tools may be particularly important in technology M&A, where the target’s value may significantly depend on proprietary models, training data, data rights, AI-generated assets, and/or AI-enabled products.
Concluding Thoughts: The More Things Change, the More They Stay the Same?
Ultimately, AI-related exclusions may become a recurring issue in RWI underwriting. RWI may still play a vital role, and standalone AI insurance products may fill certain gaps as they appear, but neither RWI nor specialized coverage should automatically be assumed to provide a complete backstop for AI risks. A strong understanding of both the target’s industry and AI risk will remain critical to negotiating AI-related exclusions to narrow, fact-specific carve-outs, rather than overly broad language that can materially erode a policy’s value. Similarly, traditional and strategically negotiated contractual risk allocation mechanisms may take on renewed importance in the AI era. What’s old may be new again.
