How To Use Undetectable AI on Claude Output
Claude 3.5 is one of the best LLMs in the market today, even outperforming GPT-4 at times. However, when it comes to writing, it can still be flagged as machine-generated from time to time. Here’s how I’ve been using Undetectable AI to fix that problem:
John Angelo Yap
Updated September 24, 2024
An AI doing his assignment, generated with Midjourney
Reading Time: 6 minutes
Claude has been around long enough that I feel like it deserves its own title as a potential ChatGPT killer.
Don’t get me wrong — ChatGPT’s pretty good (and it will continue to be), but I only ever use LLMs as my writing assistant, and I’ve always found Claude to be the better writer between the two.
Unfortunately, it can still be flagged as AI by detectors from time to time. Even articles that I wrote myself and only asked Claude to proofread can be falsely accused as machine-generated.
So, what can we do? Use AI humanizers. Here’s how I’ve been using Undetectable AI with Claude.
What is Undetectable AI?
Tired of your writing getting flagged as AI-generated? Try using Undetectable AI.
At first glance, Undetectable AI may seem like your average AI paraphrasing tool — but it’s not. It’s an AI bypassing tool or AI humanizer that can turn any text that you suspect will be flagged as machine-generated into something that resembles human writing. And that’s just their main feature, as they also have an AI human typer.
Text that gets processed by Undetectable AI will be “humanized” by removing certain AI telltale signs like repetition, common words, and adding intentional errors.
To know more about this tool, here is our full review of Undetectable AI.
What is Claude?
Claude is Anthropic’s answer to OpenAI’s ChatGPT. It's designed to engage in a wide range of tasks, from analysis and coding to creative writing and problem-solving. As a large language model (again, like ChatGPT), Claude processes and generates human-like text, allowing for natural conversations and complex reasoning.
Anthropic's latest lineup of AI models is Claude 3. The model that’s available to the public for free is Claude 3.5 Sonnet — the most advanced and versatile of the bunch.
It’s not perfect, but I find it to be better than ChatGPT in writing. Some of its limitations include not learning from chats and a knowledge cutoff date. Some also say that Claude is pretty strict because it has clear ethical boundaries and avoids harmful content.
How To Use Undetectable AI After Claude?
So, let’s start writing, or in this case, let’s have Claude and Undetectable AI do the writing for us. Here are the only three steps you need to know to use Undetectable AI after Claude.
Step #1: Ask Claude To Write For You
This one is pretty straightforward and self-explanatory: ask Claude to write for you. But I do want to dig a little deeper into building a good prompt.
Like I said earlier, Claude (and most LLMs really) has a knowledge cutoff date and can’t browse the internet for more information. That’s why apart from having clear instructions, I also highly recommend giving Claude as much information you can about the subject at hand.
Why?
Think of it this way: anyone who reads your article will use other methods of figuring out if you used AI. One of the easiest ways to determine AI writing is the presence of hallucinations or fake information — something that can be avoided with enough information.
Here’s an example of a good prompt to start:
And here’s what Claude gave me in return:
Claude Output |
The summer of 2024 brought alarming evidence of accelerating global warming. Europe sweltered under temperatures 1.54°C above average, with the Mediterranean and Eastern Europe hit hardest. Tragically, extreme heat claimed over 1,300 lives during the Hajj pilgrimage in Saudi Arabia. China experienced its hottest summer on record, while the US saw unprecedented urban heat, disrupting infrastructure. Even traditionally cooler regions felt the impact, with Australia recording its hottest August ever. The transition from El Niño to La Niña offered little respite, as heat-related fatalities mounted globally. This crisis underscored the urgent need for climate action, as rising temperatures increasingly threaten human health and safety worldwide. |
Step #2: Select the Proper Output Parameters using Undetectable AI
What makes Undetectable AI unique from other AI bypassing tools (apart from HideMyAI) is that it has output customization features that can change how the tweaked version will look like. This is divided into three parameters:
- Readability: High School, University, Doctorate, Journalist, Marketing
- Purpose: General Writing, Essay, Article, Marketing Material, Story, Cover Letter, Report, Business Material, Legal Material
- Balance Level: More Human, More Readable, Balanced
It’s important that you choose the correct parameters so that your writing comes out okay in the end. For this example, I’ll use university for readability, essay for purpose, and more human for balance level.
Original Claude Output | Undetectable AI’s Version |
The summer of 2024 brought alarming evidence of accelerating global warming. Europe sweltered under temperatures 1.54°C above average, with the Mediterranean and Eastern Europe hit hardest. Tragically, extreme heat claimed over 1,300 lives during the Hajj pilgrimage in Saudi Arabia. China experienced its hottest summer on record, while the US saw unprecedented urban heat, disrupting infrastructure. Even traditionally cooler regions felt the impact, with Australia recording its hottest August ever. The transition from El Niño to La Niña offered little respite, as heat-related fatalities mounted globally. This crisis underscored the urgent need for climate action, as rising temperatures increasingly threaten human health and safety worldwide. | In the summer of 2024 we witnessed concerning signs of climate change. Europe endured temperatures soaring 1 page above the norm with the Mediterranean and Eastern Europe bearing the brunt. Sadly more, than 1 300 lives were lost to heat during the Haj pilgrimage, in Saudi Arabia. China faced its summer in history while the US grappled with urban heatwaves that disrupted city systems. Typically cooler areas felt the effects as Australia marked its August on record. The shift, from El Niño to La Niña didn't provide relief as heat related deaths surged worldwide.This situation highlighted the pressing necessity, for climate initiatives since escalating temperatures are posing a growing risk to well being and security globally. |
So, without further ado, let’s check how well it does against Sapling (one of the best free AI detectors according to our 2024 testing)
Yep, zero. Not even a whiff of AI in that text.
At this point, we’ve already achieved our goal of turning AI-generated text into something that resembles human writing. But we’re not done yet.
Step #3: Correct Undetectable AI’s Grammar
Undetectable AI has this habit of adding intentional errors to your writing. While it’s true that it lowers AI likelihood score, it also decreases the readability of your input, which isn’t really ideal. That’s why you need to correct its grammar first, and here’s how:
- Make sure that the punctuation marks are properly placed.
- Add or remove sentence breaks to avoid monotony.
- Correct wrong grammar and spelling.
Here’s how I’d correct Undetectable AI’s output above:
Undetectable AI’s Version | Undetectable AI + Human Tweaking |
In the summer of 2024 we witnessed concerning signs of climate change. Europe endured temperatures soaring 1 page above the norm with the Mediterranean and Eastern Europe bearing the brunt. Sadly more, than 1 300 lives were lost to heat during the Haj pilgrimage, in Saudi Arabia. China faced its summer in history while the US grappled with urban heatwaves that disrupted city systems. Typically cooler areas felt the effects as Australia marked its August on record. The shift, from El Niño to La Niña didn't provide relief as heat related deaths surged worldwide.This situation highlighted the pressing necessity, for climate initiatives since escalating temperatures are posing a growing risk to well being and security globally. | The summer of 2024 brought with it alarming evidence of climate change. Europe had temperatures that were one page over average, with Eastern and Mediterranean regions suffering the worst. Regretfully, almost 1,300 people lost their lives in Saudi Arabia due to heat during the Haj. China faced the hottest summer in history while the US dealt with urban heat waves that damaged city systems. The repercussions were felt in normally colder locations as Australia saw its hottest August on record. The transition from El Niño to La Niña did not alleviate the increase in heat-related fatalities that occurred internationally. This circumstance underscored the urgent need for climate measures, as rising temperatures are increasingly endangering people's health and security on a global scale. |
Does it still pass AI detectors? The answer is yes:
Will It Pass AI Detection Tools?
Test #1
Undetectable AI and Claude successfully beats Sapling!
AI Likelihood Score: 0%
Undetectable AI and Claude successfully beats GPTZero!
AI Likelihood Score: 19%
Undetectable AI and Claude successfully beats Copyleaks!
Test #2
Undetectable AI and Claude successfully beats Sapling!
AI Likelihood Score: 18.9%
Undetectable AI and Claude successfully beats GPTZero!
AI Likelihood Score: 1%
Undetectable AI and Claude successfully beats Copyleaks!
Test #3
Undetectable AI and Claude successfully beats Sapling!
AI Likelihood Score: 16.2%
Undetectable AI and Claude successfully beats GPTZero!
AI Likelihood Score: 1%
Undetectable AI and Claude successfully beats Copyleaks!
Overall Tally
Sapling AI | GPTZero | Copyleaks | |
Test #1 | 0% | 19% | Human |
Test #2 | 18.9% | 1% | Human |
Test #3 | 16.2% | 1% | Human |
Average | 11.7% | 7% | Human (3/3) |
The Bottom Line
It doesn’t matter whether you’re using Sapling, GPTZero, or Copyleaks — Claude with Undetectable AI is just an unstoppable force when it comes to avoiding AI writing.
If you think about it, the reason for this is pretty simple. Apart from Claude’s great writing, most AI detection tools are trained using data from the most popular LLM, which happens to be ChatGPT. That means that Claude (and Gemini, for that matter) is less identifiable to AI detectors.
But of course, that’s not to underscore how much impact Undetectable AI has on the low AI likelihood scores above. My only real complaint is their over-reliance on adding errors to the input, but at the end of the day, that’s something that can easily be fixed either manually or with QuillBot.
Before I go, and I can’t stress this enough: don’t use Undetectable AI to cheat. Only use it for your personal use or to avoid getting falsely flagged as AI.
Want to Learn Even More?
If you enjoyed this article, subscribe to our free newsletter where we share tips & tricks on how to use tech & AI to grow and optimize your business, career, and life.