Ethical use / Regular use
Everyday use, without the quiet mistakes.
Six habits for work, personal projects, writing, and code. None of them slow you down much; all of them prevent the kind of mistake that is embarrassing to explain afterwards.
You are accountable for whatever you send
The model produces text that is plausible. Plausible is not the same as true, and nothing checks the difference before it reaches you. If you forward it, publish it, or ship it, it is your claim now.
This is not a reason to avoid AI. It is a reason to read what it gives you. The failure mode is not obvious rubbish — it is a confident, well-written paragraph with one wrong number in the middle.
Before you send it: could you defend every factual claim in this if someone challenged it? If not, check those parts or cut them.
Do not paste other people’s information into it
The single most common ethical mistake is casual. Someone pastes in a customer email, a colleague’s CV, a medical letter, or a client contract to get a quick summary — and hands another person’s private information to a third party without asking them.
- Never paste: anything identifying a real person — health, finances, immigration status, contact details, private messages.
- Be careful with: employer data, unreleased work, anything under an NDA. Check whether your workplace has an approved tool before using a personal account.
- Safer habit: strip the names and details first. A summary rarely needs to know who it is about.
Say so when it would change how someone reads it
You do not need to announce AI on every email. The honest test is whether the person would feel misled to find out.
Using it to tidy your grammar is like using a spellchecker — nobody expects a disclosure. Sending a condolence message, a personal recommendation, or a performance review that AI wrote is different, because the value of those was supposed to be that you wrote them.
- No disclosure needed: grammar, formatting, rewording your own draft, brainstorming.
- Say something: published or professional work substantially generated by AI, anything presented as personal, and any context where someone is relying on your expertise.
Never use it to imitate a real person
These tools are good at mimicking a specific voice, and that capability has an obvious misuse. Generating text, audio, or images that present a real person as saying something they did not is deception regardless of intent — and depending on where you live, potentially illegal.
The same applies to reviews, testimonials, and support messages written to look like they came from real customers or staff.
Know what it is not qualified to be
It will answer medical, legal, and financial questions fluently and at length, because fluent answers are what it produces. That fluency is not competence, and it has no idea what it does not know.
Use it to understand terminology, prepare questions, or make sense of a document you already have. Do not let it replace a doctor, a lawyer, or an accountant — and if you are in crisis, please talk to a real person or a crisis line rather than a chatbot.
Assume it inherited the internet’s biases
It learned from human writing, including all the stereotyping in it. That shows up quietly: in whose name gets attached to which job, whose English gets called unprofessional, which candidate gets described as a culture fit.
Be especially careful using it for anything that sorts people — hiring, admissions, lending, moderation. If you must, keep a human decision-maker who can be questioned, and check the outputs for patterns rather than assuming neutrality.