Using AI Responsibly — Security Medic Consulting

Students & Early Career / Ethics & Judgment

12th Grade · First-Year College · Job Seekers

Not a lecture. Not a list of forbidden things. A framework for making your own good decisions — because the rules are still being written, and the people writing them are watching how you use these tools.

§01 Why this is genuinely hard right now

You’re being asked to follow rules that don’t exist yet.

Here’s the honest situation. Your teachers, professors, and eventual employers are all asking you to use AI “responsibly.” But nobody — not your school, not your college, not your company, not even the people making these tools — has fully figured out what that means. They’re writing the rules in real time, partly by watching what people like you do.

That’s uncomfortable. It also means you’re not being asked to follow a clear map. You’re being asked to navigate without one.

This piece isn’t going to give you a rule list to memorize. Rule lists go stale in six months. Instead, it’s going to give you a way of thinking about AI use that will still work when specific rules change — and a few specific moves that will keep you out of real trouble in the meantime.

The reframe that makes everything else easier: using AI responsibly is not about following somebody else’s rules. It’s about being the kind of person who makes decisions you’d still be comfortable with if they were published.

The core principle

If you wouldn’t be comfortable explaining your AI use to the person whose judgment you’re affecting — the teacher grading you, the employer hiring you, the friend relying on your work — don’t do it that way.

Every specific rule that follows is a version of this one. Once you internalize it, most questions answer themselves.

§02 A decision framework you can actually use

Four questions. In order. Every time.

When you’re not sure whether your AI use is okay, walk through these four questions. They take about thirty seconds and will give you a reliable answer in almost every situation.

What is this work supposed to demonstrate?

A school assignment usually exists to show what you can think, understand, or do. A work task usually exists to produce a result. These are different. If the work is meant to measure your skill, using AI to do the thinking dodges the purpose. If the work is meant to produce a useful outcome, using AI to get there faster is often exactly right.

Would the person who receives this be surprised to learn how AI was involved?

This is the disclosure test. If your teacher would feel deceived, or your employer would feel misled, or your future spouse reading your cover letter would feel like they got catfished — that’s your answer. Surprise is the signal that disclosure was owed. If the involvement is obvious or expected, you’re probably fine.

Does the AI have access to information it shouldn’t?

This is the privacy test. Pasting a friend’s text messages, a medical record, a confidential document, or a company’s internal data into a public AI tool is a privacy violation — potentially a legal one, depending on the context. Treat anything you paste into an AI the way you’d treat anything you post on a billboard. Some of it is fine. Some of it absolutely isn’t.

Am I prepared to be responsible for the output?

This is the accountability test. The moment you submit AI-assisted work with your name on it, you own every claim, fact, citation, and consequence. If the output is wrong, you’re wrong. If the citation is fake, you’re responsible. If the code has a bug, it’s your bug. Verify before you ship.

Most of the AI mistakes that damage people’s grades, reputations, and jobs come from skipping one of these four questions. Walk through them, and the right answer usually becomes obvious.

§03 Academic integrity — what’s really going on

Your teachers aren’t against AI. Most of them are against dishonesty.

The conversation about AI and schoolwork often gets framed as “teachers don’t understand AI” or “students are cheating more than ever.” Both are oversimplified. What’s really happening is that schools are trying to preserve what education is actually for — the development of your own mind — while acknowledging that AI exists and isn’t going away.

Here’s the part that often goes unsaid: the reason assignments exist isn’t to produce essays. Essays are the artifact. The point is the thinking you do to produce one. Reading sources, wrestling with a question, forming an argument, finding the right words — that’s the muscle the assignment was supposed to build. If AI does that work, the muscle doesn’t develop. You haven’t “cheated the teacher.” You’ve cheated yourself out of the growth the assignment existed to cause.

That’s the serious version. Here’s the practical version.

Use that’s almost always okay

Asking AI to explain a concept you didn’t understand in class
Using AI as a study partner to quiz you on material
Getting feedback on a draft you wrote
Asking AI to help you understand a piece of writing (summarize, paraphrase, identify the thesis)
Using AI to help brainstorm ideas before you commit to one
Having AI check your work for basic errors (if the assignment wasn’t a spelling test)

Use that’s usually a problem

Submitting AI-generated text as your own writing
Having AI do the core thinking (thesis development, argument construction) on a paper meant to measure yours
Using AI on a test or quiz that prohibits it
Taking AI-generated citations at face value without checking that the sources exist
Using AI to “translate” your writing into a more impressive style that doesn’t actually sound like you

Use that depends on the class and the assignment

AI-assisted editing of your own writing (many teachers allow this; some don’t)
Using AI to generate example problems or practice material
AI-assisted coding on programming assignments (policies vary enormously)
AI as a research assistant for finding sources (okay in most cases, but verify what it finds)

The single biggest trap

AI can invent citations that look completely real. Fake authors. Fake journal names. Fake DOIs. They sound plausible because they were generated from patterns of real citations. If you ever submit a paper with AI-generated citations you didn’t verify, and even one of them is fabricated, you are likely facing an academic integrity hearing. This isn’t a “might happen.” It’s happening to students at every major university right now. Verify every citation.

When in doubt: ask your teacher or professor directly. Seriously. “I want to use AI for X on this assignment — is that okay?” is one of the best questions you can ask. Most teachers respect the question enormously. The ones who don’t are the ones whose policy you especially needed to know.

§04 Privacy — what happens to what you paste in

Treat AI tools like a stranger you’re shouting at across a crowded room.

When you paste something into a consumer AI tool, you generally don’t know where it goes, how long it’s kept, who can see it, or whether it’ll be used to train future models. Some services are better about this than others. Most are unclear. And the policies are changing as fast as the technology.

This means the burden of being careful falls on you. Which is uncomfortable, because most people just want to get work done.

Here’s a reasonable rule of thumb: don’t paste anything into an AI tool that you wouldn’t be comfortable seeing quoted in a news article. That’s overly conservative for some situations, but if you default to it, you won’t accidentally destroy your life.

Things that should never go into a consumer AI tool

Social Security numbers, driver’s license numbers, passport numbers. Ever. Under any circumstances.
Medical records, diagnosis information, therapy notes. Yours or anyone else’s.
Bank account numbers, credit card numbers, login credentials. Obvious, but people do it.
Other people’s private information — especially text messages, emails, or photos they sent you privately. You don’t have the right to share someone else’s private communication with an AI, even for “help writing a response.”
Workplace-confidential documents, client data, or proprietary information when you’re interning or employed. This is one of the fastest ways to get fired or sued as a new employee. Many companies have strict policies; read them.
Anything you’re legally required to keep confidential — student records, patient information, legal case details, etc.

Things that are usually okay

Public information (articles, published content, things already on the open web)
Your own non-sensitive notes and drafts
Generic coding questions that don’t involve proprietary code
Academic content you’re studying (textbook passages, study materials)
Creative writing you wrote yourself

A nuance worth knowing

Business-tier AI tools (ChatGPT Team/Enterprise, Claude for Work, Gemini for Workspace) often have stronger privacy guarantees than free consumer versions. If you’re at a company that provides one, use it. If you’re a student and your school offers an approved AI tool, use that instead of the consumer app. The protections are real and they matter.

§05 Bias, representation, and whose voice is in the machine

The model learned from us. Including the parts of us we’re not proud of.

LLMs were trained on enormous amounts of human-written text — a lot of the internet, books, articles, forums, and more. That includes everything wise humans have written, and everything biased, stereotyped, outdated, or plain wrong that humans have written. The training process tries to filter the worst of it, but no filter is perfect.

What this means in practice:

Ask an AI to generate an image of “a CEO” and you’ll often get a white man in a suit. Ask for “a nurse” and you’ll often get a white woman. These reflect patterns in training data, not reality.
Ask an AI to summarize a controversial topic and you’ll often get whichever framing dominates the internet — which isn’t necessarily the correct or most thoughtful framing.
Some AI tools have been caught generating output that’s culturally insensitive, medically biased, historically distorted, or politically skewed in one direction or another.
English-language models often work less well for other languages, dialects, and cultural contexts — sometimes badly so.

This isn’t a reason to reject AI. It’s a reason to use it with your eyes open.

What this asks of you

If you’re using AI in situations where representation matters — writing about people, making decisions about people, generating images of people, researching communities other than your own — treat the AI output as a starting draft that reflects averages, not a finished product that reflects reality. Push back on it. Specify. Ask for variety. And verify when you can.

If you’re using AI to research a topic that might be politically or socially contested, get at least one source from outside the AI conversation. LLMs can sound authoritative while reflecting only one slice of a debate. The sound of confidence is not the sound of accuracy.

The AI doesn’t have opinions. It has statistical echoes of opinions it was trained on. When it says something with confidence, that confidence is performance — not truth.

§06 Real situations, walked through

Ten common scenarios — and how to think about them.

This is the section to bookmark. These are situations that come up constantly.

Scenario 1 — College application essay

Can I use AI to help me write my Common App essay?

Heavily restricted. The essay exists precisely to show the admissions reader who you are, in your own voice. Using AI to generate the essay defeats the entire purpose, and admissions offices are getting better at spotting AI-written prose. Using AI to brainstorm topics, get feedback on a draft you wrote, or identify weak transitions is usually fine — but the actual writing needs to be yours. When in doubt: less AI, not more. The authentic voice of a 17-year-old beats a polished AI-edited version every time.

Scenario 2 — Group project

My group is letting AI write most of our paper. I’m uncomfortable. What do I do?

Raise it once, then act. Say to the group: “Before we submit this, has everyone read the professor’s AI policy?” Most of the time, this is enough — people didn’t realize, or assumed someone else had checked. If the group insists on submitting work that violates policy, you have two choices: submit your own version, or report the situation. Neither is pleasant. The wrong choice is going along with a submission you know is dishonest, because if it’s caught, you’ll be held responsible with the others.

Scenario 3 — Job application cover letter

Can I use AI to help write my cover letter for a job I really want?

Yes, with care. Most hiring managers assume AI was involved somewhere in modern applications. What they don’t want is a cover letter that sounds like a template generated for anyone applying to anywhere. Use AI for structure and polish, but keep specific details, real stories, and your actual voice. The cover letter that wins is the one where the hiring manager thinks “this person actually looked at what we do.” AI can’t do that part for you.

Scenario 4 — Technical interview

Can I use AI during a take-home coding challenge?

Depends entirely on what they said. Read the instructions carefully — companies now explicitly state their policy, and it varies. Some say “use any tools you’d use on the job, including AI.” Some say “no AI — we want to see how you think.” If it’s unclear, ask. “I want to be upfront — is AI assistance allowed on this challenge?” is a better question than making an assumption you can’t defend later. Getting caught using AI when it wasn’t allowed ends the interview process immediately.

Scenario 5 — First day at an internship

My new company hasn’t said anything about AI. Can I use ChatGPT to help with tasks?

Ask first — specifically. “Hey, I noticed we use a lot of [X tool]. Is there guidance on AI tool use?” is a perfect new-hire question. Many companies have specific approved tools and strict rules about what data can be fed into them. Using an unapproved tool, especially with company data, is one of the fastest ways for an intern to get fired. The question signals that you’re thoughtful, not just enthusiastic.

Scenario 6 — Personal writing

I’m writing a short story for fun. Can I use AI however I want?

Yes, mostly. If the work is purely for you, use whatever tools you enjoy. If you share it publicly, consider disclosing AI involvement — readers generally appreciate the honesty, and the norms around this are forming. If you submit it to a literary contest or publication, read their AI policy carefully. Many forbid AI-assisted work, or require disclosure.

Scenario 7 — Helping a friend

My friend asked me to help write a message to someone they’re dating. Can I use AI?

Worth pausing. You’re being asked to help shape your friend’s voice to another human who thinks they’re reading your friend’s actual thoughts. There’s nothing strictly “unethical” about using AI for a nudge — but using it to generate the whole message means two real people are building a relationship through a third party that neither of them signed up to involve. If you wouldn’t be comfortable with your own relationship being built on AI-written messages from a stranger, don’t help set that up for someone else.

Scenario 8 — Research paper citations

ChatGPT gave me 10 great citations for my paper. Can I use them?

Only the ones you’ve verified. AI-generated citations are the single most common cause of academic integrity violations right now. The AI doesn’t know if a paper exists — it generates plausible-sounding titles, authors, and DOIs based on patterns. Before citing anything the AI provided, find the actual source. If you can’t find it in a real database, it doesn’t exist. Don’t cite it.

Scenario 9 — Mental health

Is it okay to use AI like ChatGPT as someone to talk to when I’m stressed or down?

For light moments, sure. For real struggles, no. AI chatbots can feel surprisingly comforting, and a quick conversation to organize your thoughts is fine. But AI isn’t a therapist — it doesn’t remember you between conversations, can’t notice warning signs, doesn’t have real judgment about your safety, and has been caught giving dangerous advice in mental health crises. If you’re actually struggling, talk to a real human: a school counselor, a trusted adult, your doctor, or a crisis line. They exist for exactly this reason. An AI is not a substitute.

Scenario 10 — Creating images of people

Can I use AI image tools to create pictures of celebrities, my friends, or someone I’m interested in?

Generally no, and for some categories, never. Creating realistic AI images of identifiable people without consent is a fast-growing area of legal risk. Creating sexualized or compromising images of real people — including classmates — is illegal in many jurisdictions and is causing real harm to real people right now. If you’re using AI image tools, stick to fictional characters, stylized art, or yourself. The person whose face you’re borrowing didn’t consent, and “it was just a joke” is not a defense when the harm is done.

§07 Myths that make responsible use harder

Bad ideas that circulate freely.

Myths about AI responsibility, honestly corrected

Myth “If it’s not explicitly banned, it’s fine.”

Reality Policies can’t cover every scenario. The absence of a specific rule doesn’t mean a behavior is okay — it means the rule-writer didn’t anticipate it yet. Use the four-question framework from §02. If something would surprise or upset the person relying on your work, the absence of a written rule isn’t your shield.

Myth “AI detectors will catch anyone using AI, so it’s not worth the risk.”

Reality Mixed. AI detection tools have real false-positive rates and real false-negative rates. They catch some people. They miss others. They sometimes flag real writing as AI-generated. The decision about whether to use AI shouldn’t hinge on whether you’ll get caught — it should hinge on whether what you’re doing is the right thing regardless. The “won’t get caught” version of responsibility is the version that catches up with you later.

Myth “Everybody’s doing it.”

Reality A lot of people are using AI, yes. But “everybody’s doing it” isn’t true — it’s what people tell themselves before doing something they know wouldn’t pass their own test. The people who use AI thoughtfully don’t talk about it the same way the people who misuse it do. You’re not required to match whatever you think the average person is doing.

Myth “Disclosing AI use makes my work look worse.”

Reality Often the opposite. Smart disclosure — “I used AI for X and Y, but the thinking in sections 2 and 3 is mine” — signals self-awareness, honesty, and judgment. Employers and professors are increasingly valuing candidates who can talk clearly about their AI use. The ones who look worse are the ones who try to hide it and get caught.

Myth “AI is neutral — if it’s biased, that’s on the training data, not me.”

Reality When you use AI output, you become responsible for it. You can’t blame the tool for what you chose to pass along. If you submit biased or inaccurate content under your own name, it’s your content. The AI doesn’t take the fall. You do.

§08 Terms you’ll encounter

A responsibility & ethics glossary.

Academic integrity

The principle that the work you submit represents your own thinking and effort, with proper credit to others where used. Violations include plagiarism, unauthorized collaboration, and undisclosed AI use in situations where it wasn’t permitted.

Disclosure

Explicitly stating how AI was used in producing something. Required in many academic, professional, and creative contexts. Best practice: disclose voluntarily, even when not required.

Data minimization

The principle of sharing only the information strictly needed for a task. Applied to AI: don’t paste more than you have to. If the AI doesn’t need someone’s name, remove it.

Personally Identifiable Information (PII)

Information that can identify a specific person — name combined with address, phone number, SSN, medical data, etc. Should never be pasted into consumer AI tools.

The legal agreement you accept when you use an AI tool. Tells you what the company can do with your data, what you can do with the output, and what you’ve agreed to. Worth skimming at least once.

Training data

The text, images, or code the AI learned from. May or may not include things you paste into the tool — policies vary by provider and tier.

Hallucination

When the AI confidently produces false information. Especially common with citations, statistics, historical facts, and technical details. Always verify before relying.

Algorithmic bias

Systematic unfairness in AI output that reflects biases in training data or model design. Can show up as stereotyped image generation, skewed summaries, or disparate performance across groups.

Deepfake

AI-generated or AI-manipulated audio, video, or images made to look authentic. Creating non-consensual deepfakes of real people — especially intimate ones — is illegal in a growing number of jurisdictions.

Consent

Permission from someone to use their image, voice, writing, or data. Using AI on someone’s content or likeness without consent is often an ethical problem and increasingly a legal one.

Attribution

Giving credit for sources, ideas, or content you use. Applies to AI output too — some publishers, professors, and employers require it.

Fair use

A legal doctrine that allows limited use of copyrighted material without permission. Complex and context-specific — don’t assume AI-generated content is automatically fair use.

§09 What to carry forward

Seven things to remember when rules get fuzzy.

The core test: would you be comfortable if how you used AI was published? If yes, you’re probably fine. If no, reconsider.
Use the four-question framework. Purpose, disclosure, privacy, accountability. Thirty seconds of thinking saves a lot of regret.
Your teachers aren’t the enemy. The vast majority are trying to preserve what education is actually for. Ask them directly when you’re unsure.
Verify every citation, fact, and source the AI gives you. Hallucinations are common, and you’re responsible for what you submit.
Don’t paste private things into public tools. Especially other people’s private things. Especially at a new job.
Disclose generously, not stingily. Openness about AI use signals maturity, not weakness. The people trying to hide it are the ones who get in trouble.
You are responsible for what goes out under your name. Full stop. The AI is a tool. You are the person.

The students and employees who navigate this era well are not the ones who use AI the most, or the least. They’re the ones who use it deliberately — who can explain what they did, why they did it, and what they did themselves. Be that person.

The next card in this series gets into the career side: what jobs in AI actually exist, what to study if you’re interested, and — just as importantly — which paths don’t require you to code at all. If you’ve been wondering whether there’s a place for you in this field, that’s the piece to read next.

Next in this series

Career Pathways in AI

Roles that didn’t exist a decade ago, what to study, and how entry-level positions are evolving across tech, healthcare, government, and business. You don’t have to code to have an AI career.