As a software engineer, I believe that liability for AI-caused harm should rest more with the deployer than the developer. Developers create tools with certain specifications, but the context in which those tools are used is controlled by those deploying them. Deployer decisions and oversight significantly impact how AI systems behave in the real world.