AI Ethics in Business: Practices That Actually Hold Up
Business AI ethics is not a manifesto. It is a set of practices around disclosure, verification, and accountability. Here are the ones I follow. AI ethics in a business context gets discussed as either a compliance checkbox or a philosophical debate. Neither is useful when you are deciding whether to ship an AI-generated report to a client. The practices that actually hold up are smaller and more concrete: disclose where AI contributed, verify before you ship, and keep a human accountable for the output. Here is how I apply each, drawn from a year of using AI tools in client work where the output had real consequences. Disclose AI Use Where It Affects Trust If AI generated a meaningful part of something a client or customer reads, I say so in the place they would naturally look for how the work was produced. The line is not whether AI touched the work. AI touches a lot of work in invisible ways, like spellcheck. The line is whether a reader would feel misled if they learned after the fact. If they would, I disclose up front. This is simpler to apply than it sounds. A first draft I rewrote completely does not need disclosure. A data summary the model produced and I did not verify claim by claim does. A chatbot that speaks for the company always needs disclosure, because the reader assumes a human otherwise.