Never-share list

Quick — think of the biggest gossip you've ever known. The neighbor, the coworker, the cousin who cannot hold a secret to save their life. Got the face? Good. Now imagine handing that person your old company's strategy documents. You'd never, right? Well, here's the uncomfortable truth: every time you paste something into a public AI chat, you might be doing exactly that. These tools are wonderful at organizing your work, and terrible at keeping your secrets. So today we're talking about AI hygiene, and I'd argue it's one of the most important lessons in this whole course. Because you're using AI in your search — for your interview prep, your case studies, your resume polish — and you should be. But here's the thing: the discipline that separates a smart user from a liability is simply managing what you feed the machine. What you type in can sit in logs, and in some cases it can teach the model itself. So you treat every prompt like a postcard, not a diary. Now, from the hiring side of the desk, can I tell you why this suddenly matters so much? Two reasons. First, employers have gotten serious about tracing careless AI use, and a candidate who leaks a former employer's information doesn't read as tech-savvy — they read as a walking risk. And second, the flow runs both directions now. Their tools are collecting and analyzing your data too. So privacy isn't a side topic in your search anymore. It's a professional skill, and you're about to have it. All right, let's start with the never-share list, and I want you to treat these four as absolute. One: your own sensitive personal identifiers. That means anything that could tie you back to a data set. Two: proprietary material — internal processes, trade secrets, code that belongs to your current or former employer. Three: non-public numbers — revenue figures, growth metrics, strategy roadmaps. And four: anything covered by a confidentiality agreement, full stop. Breaking that last one isn't just a reputation problem — it can become a legal one. So how do you still get value from the tools? You anonymize, and there are four moves you'll use. Generalize: say a top-five financial technology firm, never the company's actual name. Mask: turn raw numbers into percentages — improved efficiency by twenty percent — instead of pasting internal reports. Abstract: describe the concept and the logic, never the proprietary architecture underneath. And redact: project codenames, client names, your colleagues' names — all of it comes out before you hit enter. Does that make sense? You keep the shape of the story and strip away everything that identifies the players. And when you're drafting a case study for interviews, two rules. Give the AI the framework — the problem and the solution in general terms. Let it help you with structure and flow, never with the confidential content. Or better yet, you turn the real project into a hypothetical one, because a well-built hypothetical shows your skill just as clearly and discloses nothing. Now let me tell you about a star analyst, because this story keeps me up at night. He wanted a fast summary of a fifty-page internal strategy document, so he pasted the whole thing into a public AI chat. One paste. That document held his company's most sensitive competitive plans, and when that material later surfaced outside the company, the leak was traced straight back to him. He was fired for a security breach — a genuinely talented person, undone by thirty seconds of convenience. And the part that stings? Every anonymization move you just learned would have saved him. Now, the honest part, because I don't want you walking away scared of the tools. Use them — absolutely, freely, daily — just with clean inputs every time. This lesson isn't anti-AI, it's pro-you. And one more piece of awareness for the road: when you're the candidate, employers' systems may analyze your video interviews and assessments, and you'll sometimes be asked to consent to that. Read what you're agreeing to before you ever click accept. The rules around disclosure and consent are tightening in a lot of places, which is good news for you — but the first line of defense is you actually reading the screen. So here's your assignment tonight, and it's a two-minute drill. Open the AI tool you use most and look at your last five prompts. Any company names? Real numbers? Project codenames? Scrub them, and build yourself the placeholder habit — company X, proprietary project — so that the clean version becomes your default. Then take five bullets from your resume and practice the same scrub, because the discipline transfers everywhere you write. Go scrub those prompts.