I believe that liability for AI-caused harm should primarily fall on the deployer. Developers create the tools, but the deployers decide the context and purpose of their use. It's crucial to ensure that those who implement AI systems are held responsible for their applications, as they control the environment and parameters within which the AI operates.