Artificial intelligence (AI) is moving quickly from internal productivity use cases into customer-facing functions that answer questions, provide guidance, and increasingly shape decisions. As that shift continues, AI is becoming more than a technology issue; it is emerging as a meaningful source of corporate liability. Early litigation suggests that the exposure may extend well beyond privacy and cybersecurity concerns to include real-world harms tied to what AI systems communicate, recommend, or enable.

Companies May Be Liable for AI Outputs

One early theme is already clear: companies may have limited ability to distance themselves from the outputs of their AI tools. In Moffatt v. Air Canada, a 2024 decision from the British Columbia Civil Resolution Tribunal, a chatbot provided inaccurate information about bereavement fare eligibility, and the tribunal held the airline responsible for the resulting loss. The significance of that decision was not the dollar amount at issue, but the underlying principle: when AI communicates with customers, courts may treat those statements as the company’s own representations.

If that reasoning gains traction, risk managers will need to reconsider how AI changes an organization’s exposure profile. Dynamically generated outputs do not undergo the same review and judgment as traditional corporate communications, yet they may still be treated as authoritative statements made on the company’s behalf. That raises difficult questions about governance, controls, and accountability, and suggests that disclaimers alone may offer only limited protection when a public-facing AI tool causes harm.

The Larger Concern: Rising Severity

The larger concern may be claim severity. Recent allegations have moved beyond economic loss into matters involving bodily injury and wrongful death. A 2026 California lawsuit, for example, alleges that an AI chatbot provided guidance about combining substances in a way that contributed to a fatal overdose. While those allegations remain unproven, they illustrate how plaintiffs may seek to frame AI systems not merely as information tools, but as products whose defective or unsafe outputs can generate substantial damages.

For companies deploying public-facing AI, that shift could materially alter the nature of their exposure. What might once have been viewed primarily as an errors and omissions issue could begin to implicate casualty lines, product liability theories, and more complex coverage questions. As severity rises, companies may need to evaluate whether existing policy language, underwriting assumptions, and risk controls are equipped to address claims arising from AI-enabled interactions.

The Bottom Line

AI offers meaningful benefits for companies, but it is also accelerating existing liability concerns in ways that may be harder to predict and control. That shift matters because it could increase both the frequency and severity of claims while exposing potential gaps between emerging loss scenarios and existing insurance frameworks. For organizations investing in AI, the key question is no longer just what the technology can do, but how the resulting enterprise risk should be governed, insured, and monitored as claims activity evolves.