Businesses are deploying AI agents for tasks like browsing and purchasing, but a fundamental legal question about liability when these agents cause harm is moving from conference rooms to courtrooms, according to a recent analysis by attorneys at Frankfurt Kurnit Klein & Selz. The firm identifies four legal frameworks for assigning liability. First, agency law may hold the company or individual that deployed the agent responsible, as seen in a case against Workday where a court allowed discrimination claims to proceed on the theory that its AI screening tool functioned as an agent of employers. Second, product liability and negligence claims could allege defective design or lack of safeguards, a theory Florida’s attorney general has advanced in a lawsuit against OpenAI. Third, contracts between vendors and businesses are being negotiated, but the firm notes these allocations do not bind consumers or regulators. Fourth, legislators are acting, with California’s AB 316, effective January 1, 2026, barring defendants from arguing the AI itself caused harm, and the European Union’s AI Act assigning accountability to specific supply chain actors. A federal court this March granted a preliminary injunction against Perplexity AI’s shopping agent under the Computer Fraud and Abuse Act.