In September, members of the KCIC team attended and presented at Perrin’s National Asbestos Litigation Conference in Charleston, S.C. One of the most illuminating panels focused on the evolving role of artificial intelligence (AI) in asbestos litigation, particularly how in-house counsel are beginning to incorporate it into their work. While the presence of AI in the asbestos litigation space is still modest, the conversation offered a clear window into how firms are navigating both the opportunities and risks.

Today, AI is being applied to relatively low-stakes, peripheral roles such as organizing, prioritizing, and drafting responses to emails. While making these daily tasks more efficient is beneficial, they fall short of the sweeping changes some have predicted that the usage of AI would bring.

A cautionary story shared during the session illustrated why: When a generative AI tool was used to assist with legal research, it fabricated lawsuits and cited them as precedent. A Maryland lawyer was fined in 2024 for citing, in court pleadings, fictious cases produced by an AI tool.[1] This kind of “hallucination” — when an AI produces seemingly legitimate but false information — underscores the risks of using open generative tools without proper controls or oversight.

In an earlier post, we highlighted the importance of recognizing hallucination risks and maintaining confidentiality when using public models. The panel’s discussion echoed this point and encouraged all firms to focus on accuracy, data security, and validation when guiding AI adoption.

A major takeaway from the September session was the recommendation to adopt closed AI systems: tools confined to an organization’s own internal and vetted data. Closed systems don’t pull from external sources, which helps to prevent hallucinations and misattributions. Instead, they rely on content or data the firm has already reviewed and authorized. The result: more reliable outputs grounded in accurate material, with tighter oversight of validation, citation, and confidentiality. This can also be used to keep AI within a specific department or project of an organization.

As artificial intelligence evolves, its usage in organizations must be constantly monitored and evaluated. We enjoyed attending this panel at Perrin Charleston and look forward to continued conversations about this topic — as well as how KCIC can best assist those at the intersection of technology and law.

[1] Massachusetts Lawyer Sanctioned for AI-Generated Fictitious Case Citations | Maryland State Bar Association