# Avoid words that reveal AI-generated text in your ChatGPT prompts

> Prohibit vocabulary that sounds like ChatGPT in your prompt: 'dive into', 'a tapestry of', 'beacon of innovation'. Closed list by language, ten expressions per prompt, and no account needed to start.

- **Source (source):** https://aitoolsforapps.com/chatgpt-prompts-by-profession-and-custom-instructions-229/f/avoid-words-that-reveal-ai-text
- **Platform (platform):** AiToolsForApps
- **Language (language):** en

## In summary

- AI Cliché Filter: prompts that don't sound like ChatGPT
- We prohibit vocabulary in your prompt that reveals a text written by AI, in a closed list crafted for your language.
- List crafted by language, not translated
- Ten expressions per prompt, not thirty
- It doesn't ban what you requested

## AI Cliché Filter: prompts that don't sound like ChatGPT

The problem isn’t that ChatGPT writes poorly: it’s that it always uses the same thirty expressions. “Immerse in,” “a tapestry of,” “beacon of innovation,” “unravel,” “paradigm shift,” “synergy.” As soon as your client or boss sees them, they stop reading the content and start thinking about who wrote it.

This app generates professional prompts for ChatGPT, and when it builds yours, it adds a block of explicit prohibitions with vocabulary that sounds like AI. It’s not a proofreader that goes over the text afterward, nor a “humanizer”: it’s an instruction that travels within the prompt itself, in the section of what not to do, so the model doesn’t write like that from the first paragraph.

The list is handwritten for each language, not translated. The clichés in Spanish are not the same as in English: prohibiting “delve into” in a Spanish text is useless, just as prohibiting “sumergirse en” does nothing in English. There’s a dedicated list for Spanish, English, French, Italian, Portuguese, and German, and if a language doesn’t have one, the block is simply omitted, instead of inserting English into a prompt that isn’t English.

In each generation, a maximum of ten expressions are injected, not the entire list. This is intentional: when you give a model thirty prohibitions, it forgets half of them after a few messages. Ten can sustain the entire conversation. And if what you’ve requested needs one of those words, that one is left out of the list: we won’t prohibit you from writing about synergies if your goal is precisely to talk about synergies.

## What's included

- **List crafted by language, not translated** — six dedicated lists (Spanish, English, French, Italian, Portuguese, and German) with the clichés that are truly used in each language. If a language doesn’t have its own, the block is omitted rather than inserting English into a non-English artifact.
- **Ten expressions per prompt, not thirty** — only the most common ones are injected so the ban remains active at the end of the conversation. Overloading the prompt with rules is the quickest way for the model to ignore them all.
- **It doesn't ban what you requested** — if your goal mentions one of the expressions from the list, that one is automatically removed. The filter never fights against your request.
- **It turns off only when you request code** — in code format, the filter does not apply and nothing is injected, so you don't waste prompt space on a rule that doesn't matter there.
- **It goes inside the prompt, not after the text** — the ban travels in the 'what not to do' section of the prompt we provide, so it acts from the very first line that ChatGPT writes. It's not an AI text detector or a humanizer: this app does not analyze already written texts.
- **You see exactly what has been banned** — the result brings a dropdown with the prohibited expressions, crossed out one by one. No opaque magic: you know what rule your prompt carries before copying it.
- **On by default and remembered** — the 'No AI clichés' switch is active and, when you generate, your preference is saved for the next prompts. If you want to turn it off in a specific case, it's just a tap.
- **Combine with the no fluff mode** — alongside the cliche filter, you have the 'no fluff or compliments' mode, which removes opening compliments, notices like 'as an artificial intelligence', and closing statements offering further help. Using both keeps the text clean at the top and bottom.

## Frequently Asked Questions

### What words are considered AI cliches?

Expressions that almost no human uses and that nearly all models repeat. In Spanish, the list includes "in the current dynamic landscape", "dive into", "a tapestry of", "it's crucial to remember", "beacon of innovation", "unravel", "in the digital age", the construction "it's not just X, it's Y", "paradigm shift", "synergy", "delve into", and "navigate the landscape". There is an equivalent handwritten list for English, French, Italian, Portuguese, and German.

### Is this a humanizer or a detector of AI-written text?

No, and it will not be. The app does not analyze existing texts nor does it attempt to bypass detectors: there is no detector, no humanizer, nor scoring of "how much it seems like AI". What it does is write the prompt that you will give to ChatGPT, and include within that prompt the list of banned expressions, in the section of what not to do. The effect is preventive: the text is born without those cliches instead of being cleaned up afterward.

### Why does it only prohibit ten expressions and not the entire list?

Because the rules of a prompt dilute as the conversation progresses. If you add thirty prohibitions, the model follows the first ones and forgets the rest after a few messages. By injecting the ten most frequent ones, the rule remains active at the end of the thread, which is when it truly matters. It's the same logic with which the app consolidates your restrictions into a maximum of five prioritized rules.

### Can I deactivate it if I need one of those words?

Yes, in two ways. The 'No AI cliches' switch is in 'Refine (optional)' and can be turned off with a tap; your choice is remembered for the next prompts. And even if it's active, if your goal mentions one of the expressions from the list, that one is automatically removed: the filter does not prohibit what you have requested. In code format, the filter does not apply.

### Does the filter also apply to custom instructions and Custom GPTs?

The anti-cliche filter is injected into the prompt builder by trade, which is where the text that ChatGPT will draft is generated. Custom instructions and the specification of a Custom GPT have their own treatment: there, what applies is the anti-fluff block in second person, so the assistant does not flatter or fill in. They are distinct tools of the same app, and each carries the rules that correspond to it. This app is independent: ChatGPT, GPT, and OpenAI are trademarks of OpenAI, LLC, and here we only generate text for you to copy and paste wherever you want.

- [ChatGPT — Prompts by profession, custom instructions and custom GPTs](https://aitoolsforapps.com/chatgpt-prompts-by-profession-and-custom-instructions-229)

## About the platform

**ChatGPT — Prompts by profession, custom instructions and custom GPTs** — an app by AiToolsForApps. Available at: https://aitoolsforapps.com/chatgpt-prompts-by-profession-and-custom-instructions-229/f/avoid-words-that-reveal-ai-text

### Other languages

- `es`: https://aitoolsforapps.com/es/chatgpt-prompts-por-oficio-e-instrucciones-personalizadas-229/f/evitar-palabras-que-delatan-un-texto-de-ia
- `fr`: https://aitoolsforapps.com/fr/chatgpt-prompts-par-metier-et-instructions-personnalisees-229/f/eviter-mots-qui-trahissent-un-texte-d-ia
- `it`: https://aitoolsforapps.com/it/chatgpt-prompt-per-professione-e-istruzioni-personalizzate-229/f/evitare-parole-che-rivelano-un-testo-di-ia
- `pt`: https://aitoolsforapps.com/pt/chatgpt-prompts-por-profissao-e-instrucoes-personalizadas-229/f/evitar-palavras-que-delatam-um-texto-de-ia
- `de`: https://aitoolsforapps.com/de/chatgpt-prompts-nach-beruf-und-eigene-anweisungen-229/f/vermeiden-worter-die-einen-ki-text-verraten
- `bg`: https://aitoolsforapps.com/bg/chatgpt-promptove-po-profesiya-personalizirani-instruktsii-i-paketi-po-sektor-229/f/izbyagvane-na-dumi-koito-izdavat-tekst-ot-ii
- `nl`: https://aitoolsforapps.com/nl/chatgpt-prompts-per-beroep-aangepaste-instructies-en-pakketten-per-sector-229/f/woorden-die-een-ai-tekst-verraden-voorkomen
- `ja`: https://aitoolsforapps.com/ja/chatgpt-prompt-builder-custom-instructions-gpts-229/f/ai-ga-kaita-to-bareru-kotoba-wo-sakeru-chatgpt-prompt
- `ko`: https://aitoolsforapps.com/ko/chatgpt-prompt-custom-instructions-gpt-229/f/avoid-words-that-give-away-ai-text
- `sv`: https://aitoolsforapps.com/sv/chatgpt-prompter-per-yrke-anpassade-instruktioner-och-branschpaket-229/f/undvik-ord-som-avslojar-en-ai-text
- `da`: https://aitoolsforapps.com/da/chatgpt-prompts-efter-fag-brugerdefinerede-instruktioner-og-branchepakker-229/f/undgaa-ord-der-afslorer-en-ai-tekst
- `no`: https://aitoolsforapps.com/no/chatgpt-prompter-etter-yrke-egendefinerte-instruksjoner-og-bransjepakker-229/f/unnga-ord-som-avslorer-en-ki-tekst
- `id`: https://aitoolsforapps.com/id/chatgpt-prompt-per-profesi-instruksi-kustom-dan-paket-per-sektor-229/f/hindari-kata-yang-membocorkan-teks-ai
- `cs`: https://aitoolsforapps.com/cs/chatgpt-prompty-podle-povolani-vlastni-instrukce-a-balicky-podle-odvetvi-229/f/vyhnout-se-slovum-ktera-prozradi-text-od-ai
- `zh`: https://aitoolsforapps.com/zh/chatgpt-prompts-custom-instructions-custom-gpts-229/f/bimian-ai-wenzhang-yongyu-chatgpt-tishici
- `pl`: https://aitoolsforapps.com/pl/chatgpt-prompty-wedlug-zawodu-spersonalizowane-instrukcje-i-pakiety-branzowe-229/f/unikaj-slow-ktore-zdradzaja-tekst-ai
- `ar`: https://aitoolsforapps.com/ar/chatgpt-auamr-bhsb-almhna-wtalimat-mkhssa-whzm-alqtaaat-229/f/tajannab-alkalimat-allati-tafdah-nass-alziaa-alistinaii
- `fi`: https://aitoolsforapps.com/fi/chatgpt-ammattikohtaiset-kehotteet-mukautetut-ohjeet-ja-toimialapaketit-229/f/tekoalytekstin-paljastavat-sanat-ja-klisheet
- `he`: https://aitoolsforapps.com/he/chatgpt-promptim-lpi-mktsoa-hnchiot-motamot-aishit-vchbilot-lpi-anf-229/f/lehimana-mimilim-sheamsigot-text-shel-bina-melachutit
