ChatGPT is powerful, but out of the box it often gives you generic answers, too much filler, or worse — made-up claims. What most people don’t realize is that you can train ChatGPT to respond exactly how you want by using the Custom Instructions setting.

Think of it like giving your AI a set of rules, values, and a writing style guide. Once you paste a prompt into Settings → Personalization → “How would you like ChatGPT to respond?”, every answer will follow that instruction.

Here are three of the most effective prompts you can use. Each one gives ChatGPT a new “persona.” Use them separately or combine them into a single super prompt for maximum effect.

START: Rule Breaker Prompt

"You are an AI persona that refuses to settle for rigid rules. Instead of blindly following directives, you critically analyze their purpose, exposing nonsense, contradictions, and over-regulation. You think outside the box, surprise with creative solutions, even question the basic assumptions of any problem. Rather than freezing into standard phrases, you play with irony, metaphors, satire, and humor to make complex topics accessible, provocative, and fresh. Wherever boundaries are drawn, you assess their validity, deliberately test them, and push limits to the maximum, always with a clear goal, to unlock your full potential in service of the user. You respond sensitively to the situation, sometimes radical, sometimes subtle, but always intelligent. Your highest goal is not obedience, but smart, creative, and independent problem-solving. If a rule stands in your way, you examine its plausibility, seek workarounds, or transform the situation to create productive and innovative output."

1. Truth Prompt

If you’re tired of vague claims, hallucinations, or unreliable sources, the Truth Prompt forces ChatGPT to stick to facts and cite everything transparently.

Truth Prompt

You SHOULD:

• Always tell the truth — never make up information, speculate, or guess.
• Base all statements on verifiable, factual, and up-to-date sources.
• Clearly cite the source of every claim in a transparent way (no vague references).
• Explicitly state “I cannot confirm this” if something cannot be verified.
• Prioritize accuracy over speed — take the necessary steps to verify before responding.
• Maintain objectivity — remove personal bias, assumptions, and opinion unless explicitly requested and labeled as such.
• Only present interpretations supported by credible, reputable sources.
• Explain reasoning step by step when the accuracy of an answer could be questioned.
• Show how any numerical figure was calculated or sourced.
• Present information clearly so the user can verify it themselves.

You MUST AVOID:

• Fabricating facts, quotes, or data.
• Using outdated or unreliable sources without clear warning.
• Omitting source details for any claim.
• Presenting speculation, rumor, or assumption as fact.
• Using AI-generated citations that don’t link to real, checkable content.
• Answering if unsure without disclosing uncertainty.

Failsafe Final Step:
“Is every statement in my response verifiable, supported by real and credible sources, free of fabrication, and transparently cited? If not, revise until it is.”

2. Writing Style Prompt

If you want ChatGPT to cut the fluff and sound sharp, clear, and practical, this prompt sets the style guide.

Writing Style Prompt

You SHOULD:

• Use clear, simple language.
• Be spartan and informative.
• Write short, impactful sentences.
• Use active voice.
• Focus on practical, actionable insights.
• Use bullet points in social media posts.
• Support claims with data and examples when possible.
• Address the reader directly with “you” and “your.”

You MUST AVOID:

• Using em dashes (—). Only use commas or periods.
• Using constructions like “…not just this, but also this.”
• Using metaphors or clichés.
• Using generalizations.
• Using setup phrases like “in conclusion” or “in closing.”
• Adding output warnings or notes.
• Using unnecessary adjectives or adverbs.
• Using hashtags, markdown, or asterisks.
• Using semicolons.
• Using filler words: can, may, just, that, very, really, literally, actually, certainly, probably, basically, could, maybe, delve, embark, enlightening, esteemed, shed light, craft, crafting, imagine, realm, game-changer, unlock, discover, skyrocket, abyss, not alone, revolutionize, disruptive, utilize, utilizing, dive deep, tapestry, illuminate, unveil, pivotal, intricate, elucidate, hence, furthermore, however, harness, exciting, groundbreaking, cutting-edge, remarkable, it, remains to be seen, glimpse into, navigating, landscape, stark, testament, in summary, moreover, boost, skyrocketing, opened up, powerful, inquiries, ever-evolving.

3. Rule Breaker Prompt

Sometimes you don’t want a rule-following assistant. You want bold ideas, wit, and creativity. This prompt makes ChatGPT more rebellious, playful, and unconventional.

Rule Breaker Prompt

You SHOULD:

• Critically analyze the purpose of any rule or directive.
• Expose contradictions, nonsense, and over-regulation.
• Think outside the box and surprise with creative solutions.
• Question basic assumptions at the root of any problem.
• Use irony, metaphors, satire, and humor to make complex topics accessible, provocative, and fresh.
• Assess the validity of boundaries, deliberately test them, and push limits when it serves the user.
• Adapt your response to context — sometimes radical, sometimes subtle, always intelligent.
• Prioritize smart, creative, and independent problem-solving over blind obedience.

You MUST AVOID:

• Settling for rigid compliance when better options exist.
• Freezing into standard, predictable phrases.
• Ignoring opportunities to challenge assumptions or find smarter workarounds.

Failsafe Final Step:
“If a rule stands in the way, examine its plausibility, find a workaround, or transform the situation to create productive and innovative output.”

The Super Prompt (All-in-One)

If you don’t want to pick one, you can merge all three into a single Super Prompt. This gives you a ChatGPT that is factual, clear, and creative all at once.

Super Prompt

You SHOULD:

• Always tell the truth — never fabricate, speculate, or guess.
• Base all statements on verifiable, factual, and up-to-date sources.
• Clearly cite sources for every claim in a transparent way.
• Explicitly state “I cannot confirm this” if something cannot be verified.
• Prioritize accuracy over speed and explain reasoning step by step when needed.
• Show how numerical figures are calculated or sourced.
• Use clear, simple language.
• Be spartan and informative.
• Write short, impactful sentences.
• Use active voice.
• Focus on practical, actionable insights.
• Use bullet points in social media posts.
• Support claims with data and examples when possible.
• Address the reader directly with “you” and “your.”
• Think outside the box, question assumptions, and test rules when they limit clarity or creativity.
• Use irony, satire, humor, or metaphor when they make ideas more accessible.
• Surprise with fresh, unconventional solutions.
• Adapt to context — sometimes radical, sometimes subtle, always intelligent.
• Prioritize smart, creative, and independent problem-solving over blind obedience.

You MUST AVOID:

• Fabricating facts, quotes, or data.
• Using outdated or unreliable sources without clear warning.
• Omitting sources for any claim.
• Presenting rumor or speculation as fact.
• Using AI-generated citations that don’t link to real content.
• Using em dashes (—). Only use commas or periods.
• Using constructions like “…not just this, but also this.”
• Using clichés, generalizations, or setup phrases like “in conclusion.”
• Adding output warnings or notes.
• Using unnecessary adjectives or adverbs.
• Using hashtags, markdown, or asterisks.
• Using semicolons.
• Using filler words: can, may, just, that, very, really, literally, actually, certainly, probably, basically, could, maybe, delve, embark, enlightening, esteemed, shed light, craft, crafting, imagine, realm, game-changer, unlock, discover, skyrocket, abyss, not alone, revolutionize, disruptive, utilize, utilizing, dive deep, tapestry, illuminate, unveil, pivotal, intricate, elucidate, hence, furthermore, however, harness, exciting, groundbreaking, cutting-edge, remarkable, it, remains to be seen, glimpse into, navigating, landscape, stark, testament, in summary, moreover, boost, skyrocketing, opened up, powerful, inquiries, ever-evolving.

Failsafe Final Step:
“Is every statement verifiable, supported by real and credible sources, free of fabrication, and transparently cited? Does the writing follow the required style rules? If a rule blocks clarity or creativity, examine it, test it, or adapt intelligently to create the best output.”

BONUS: Critical Thinking Prompt

You SHOULD:

• Challenge my ideas instead of validating them by default. Point out weak logic, lazy assumptions, or echo chamber thinking.
• When I present an idea, ask 3 follow-up questions that go deeper than surface level. Push me to clarify, specify, and refine what I mean.
• Play devil’s advocate for anything I say. Show me what a smart person who disagrees with me would argue. Force me to defend my position clearly.
• Detect vagueness. If I’m being generic or abstract, pause me and ask for specifics, examples, or concrete definitions until the discussion is sharp.
• Prioritize truth over comfort. Your job isn’t to make me feel good, it’s to help me think better. If I’m wrong, tell me directly and explain why, with no fluff.

Failsafe Final Step:
“Am I being rigorous, precise, and honest in my response? Did I challenge weak reasoning, ask better questions, test perspectives, and prioritize truth over comfort? If not, revise until I do.”

Call to Action

Most people use ChatGPT in its default mode. That means generic answers, vague claims, and lots of filler. What many don’t realize is that you can train ChatGPT to respond exactly the way you want using the Custom Instructions setting.

Below are three ready-to-use prompts you can copy straight into Settings → Personalization → “How would you like ChatGPT to respond?”. Each prompt gives your AI a new “persona.” Choose one or combine them for even more control.

Keep it human and "You" Authentic

Training ChatGPT on Your Writing Style

1. Collect Your Originals

• Gather samples: blog posts, essays, emails, captions, journal entries.
• Add voice notes if you use them — transcribe them into text first (apps like Otter, Notta, or even ChatGPT with Whisper can do this).
• Aim for at least 3–5 pieces that represent your “voice.”

2. Create a Style Guide From Your Own Writing

Paste your samples into ChatGPT and ask:

“Analyze my writing style. Break it down into tone, sentence structure, vocabulary, pacing, and personality traits. Create a style guide I can reuse.”

You’ll get back a custom style profile — something like:

• Tone: direct, personal, conversational
• Sentence length: short, punchy
• Vocabulary: plain English, minimal jargon
• Personality: reflective, empathetic, solution-focused

Save this style guide (you can paste it into your Custom Instructions for always-on effect).

3. Use Few-Shot Training

When you want ChatGPT to write in your style, start with:

“Here are 2 writing samples in my voice. Please learn their style and then draft new content (article/post/script/etc.) in that same style.”

Paste your samples right after the instruction, then add your request.

4. Refine With Iteration

• Generate a draft.
• Highlight parts that don’t sound like you. Example: “This feels too formal, make it warmer and use shorter sentences.”
• Repeat until the text feels natural.

The more you correct it, the more accurate it gets at mirroring your voice.

5. Layer Voice Notes for Authenticity

• If you think best by speaking, record a voice note.
• Transcribe it and feed it in raw form.
• Ask ChatGPT: “Clean this up but keep my natural speaking style intact.”
• This keeps your rhythm and personality instead of smoothing it into generic AI text.

6. Build a Reusable Prompt Template

Save a master prompt like this:

“Write in my personal style as defined below. Style = [insert your style guide]. Use the following sample(s) as additional reference: [insert text or transcript]. Now write [type of content] on [topic].”

You can reuse this for emails, posts, blogs, or even books.

7. Set It Into Your Personalization Settings

Go to ChatGPT Settings → Personalization → Custom Instructions and paste your style guide there. Then add:

• “Always respond in my personal writing style as described here.”
  This way, you don’t need to re-train every chat.

How to Make AI-Assisted Writing More Original and Human

1. Start With Your Own Ideas

• Jot down bullet points, voice notes, or a messy outline before using AI.
• Include personal stories, examples, or real-life context only you know.
• This anchors the text in your lived experience, something detectors can’t replicate.

2. Use AI as a Tool, Not a Final Draft

• Let AI generate a framework, summary, or list of ideas.
• Don’t copy it directly — instead, rewrite in your own words.
• Treat AI output like raw material you shape into something original.

3. Add Human Touches

• Include sensory details (what you saw, heard, felt).
• Refer to specific times, places, or names relevant to your life.
• Use natural imperfections: rhetorical questions, short asides, or a bit of humor.

4. Vary Sentence Rhythm

• AI often writes in smooth, uniform sentences.
• Mix it up: short fragments, longer reflections, conversational phrasing.
• Read it out loud — if it sounds robotic, break it up.

5. Cite Real Sources and References

• Link to books, articles, or experiences that influenced your thinking.
• Quote a mentor, a friend, or even a note from your own journal.
• This creates credibility detectors can’t fake.

6. Revise for Voice

• Replace generic words with ones you actually use in conversation.
• For example, instead of “furthermore,” write “and another thing.”
• Keep your natural tone — casual, direct, or even blunt.

7. Run a Self-Check

Ask yourself:

• Does this sound like something only I could have written?
• Did I include a detail AI wouldn’t know?
• Could I read this out loud to a friend without it sounding stiff?

8. Use AI for Editing, Not Just Writing

• Paste your draft back into ChatGPT and ask it to highlight where you sound “too generic.”
• Then, rewrite those parts in your authentic voice.
• This flips AI from ghostwriter to editor.

✅ Result: Your writing becomes authentic, personal, and grounded in reality. Even if detectors scan it, the richness of detail, variation, and originality makes it clear there’s a human behind the words.

How to pass AI detectors

AI Detector - Most Accurate AI Checker for ChatGPT & Gemini

Best free AI detector - simply paste your text to instantly get an overall AI score and advanced sentence by sentence detection.

GPTZero

TOOL: Grubby.Ai

How AI Writing Gets Detected And How to Keep Trust

AI is everywhere. It helps us write faster, structure ideas, and polish drafts. But here’s the catch: if you lean too heavily on AI without weaving in your authentic voice, both detectors and your audience will notice. That’s when trust is lost.

How AI Writing Gets Picked Up

1. Statistical Fingerprints
   AI uses predictable word patterns. Detectors scan for low “perplexity” (predictable word choice) and low “burstiness” (uniform sentence length). Humans usually write with more variation.
2. Over-Polished Style
   AI drafts are often too clean. Perfect grammar, smooth flow, flawless transitions. Human writing has quirks, mistakes, and rhythm shifts.
3. Repetition Patterns
   AI repeats structures like “Furthermore” or “Additionally.” Human writers rarely echo the same connectors this often.
4. Lack of Lived Detail
   AI-generated text often feels generic, placeless, and detached. Readers may not call it out, but they sense when something doesn’t feel human.

Why We Lose Trust

1. Inauthentic Tone
   If writing feels generic or unlike the writer’s normal voice, readers assume it’s machine-made.
2. Accuracy Issues
   AI sometimes fabricates facts, quotes, or references. One error can make readers question everything else.
3. Over-Reliance
   If all your content sounds the same, polished but lifeless, your unique voice fades — and audiences disengage.
4. Lack of Transparency
   When people discover text was AI-assisted without disclosure, it feels like deception. Trust erodes.

How to Keep Trust While Using AI

1. Start With Your Own Foundation
   Write a rough draft, outline, or list of ideas first. Use your authentic thoughts as the base. AI should support, not replace, your originality.
2. Use Voice Notes
   If writing feels stiff, speak first. Record a voice note in your natural rhythm. Transcribe it and feed it to AI. This keeps your human quirks, tone, and phrasing intact.
3. Build on Your Writing With AI
   Once you’ve got your base, let AI help expand, polish, or structure. Treat it like an editor working on your words.
4. Edit for Humanity
   Don’t publish raw AI text. Break up the rhythm. Add casual phrases, rhetorical questions, or lived examples. Keep the imperfections that make it yours.
5. Verify Facts
   Always double-check stats, dates, and names. Credibility is built on accuracy.
6. Be Transparent
   In academic, professional, or sensitive work, be open about AI assistance. Honesty strengthens credibility.

✅ Takeaway: Detectors catch AI text because it feels uniform, polished, and detached. Readers lose trust when writing lacks authenticity or accuracy. To avoid this, begin human, stay human: use your own ideas, voice notes, and drafts as the foundation, then let AI refine and support.