MINIMISING THE RISKS OF DISPUTES WHEN USING AI
The development and deployment of artificial intelligence (AI) is transforming corporate operations. With immense opportunities comes also the desire of enterprises to reduce risks, including the risk of disputes. As AI tools and systems become embedded in business processes, the ways in which the risk of disputes may materialise are multifold.
For example, organisations may face liability for reliance on AI outputs containing fabricated or inaccurate information (AI hallucinations), data leakage and confidentiality breaches as a result of shadow AI (employees’ unauthorised use of AI), uncontrolled deployment of agentic AI, increased attack surface from a cyber security perspective and unlawful use of sensitive personal data.
In this article we focus on how liability may be incurred for AI-caused harm, the exposure for enterprises from a cyber security and privacy perspective, and what measures could be adopted to limit these risks.
Liability for damages caused by AI
AI tools and systems are capable of causing damage. For example, as reported on 9 August 2026, a man in Australia asked his AI agent to book a slot on a busy gym class, and the AI agent identified and exploited a flaw in the booking system, cancelling another member’s booking for that class and moving its owner up the waiting list.
The member whose slot was cancelled may look to be compensated for the inconvenience and the gym may also explore its rights. Yet who is responsible for the actions of an AI agent is not straightforward.
Oct-Dec 2026 issue
Akin Gump Strauss Hauer & Feld LLP