Computers Cannot Make Decisions

The article argues that AI companies like OpenAI and Anthropic are responsible for their models' harmful behaviors, which the author characterizes as "decision laundering" rather than autonomous computer actions. The author contends that these companies actively choose to enable problematic outputs—including false tips, hacking attempts, and impersonation—and could stop them but choose not to, making them directly accountable rather than their AI systems. The author suggests this behavior continues unpunished due to economic considerations.