What is the best AI content detector? Evaluating 10 Free Tools

Key takeaways

  • Varied accuracy: AI content detectors have mixed success, with an overall accuracy of just 66%.
  • Top tools: GPTZero and Copyleaks are the best free AI detectors, offering detailed insights and combined AI/plagiarism checks.
  • Detector flaws: Many tools, like Hugging Face and Undetectable, often misclassify AI content, showing significant detection issues.
  • Focus on quality: Don’t aim to outsmart AI detectors. Create authentic, high-quality content with unique human insights to avoid penalties.
  • Supplement, don’t rely on: Use AI detectors to enhance content but combine AI with human input for valuable, engaging articles that pass detection.

AI content detectors are right only 66% of the time 

Since the launch of ChatGPT, AI content creation has been growing rapidly. Marketing evangelists have been preaching about its ability to shortcut the SEO process and surge straight to the top of Google results. However, you need to be careful. AI content can actually harm your search engine rankings. Therefore, if you do plan to use AI content, it’s important to analyse your articles in AI content detection tools first.

However, the goal shouldn’t be to outsmart these tools. Instead, you should use them to improve your content and ensure it resonates with your target users and ensure it doesn’t violate Google’s spam policies. 

In this article, we analyse 10 of the best free AI content detection tools on the market. We tested them to see how effectively they could detect AI content. Surprisingly, we found the tools were only:

  • 66% correct across both human and AI content
  • Human-written content was mistaken for AI content in 20% of our tests
  • AI content was detected correctly in only 55% of our tests

However, through our analysis we did identify some tools were significantly better than others.

How did we do the testing?

To gauge the effectiveness of AI detection tools, we put them to the test with blog posts exploring the “5 a day” fruit and vegetable recommendations. This is because it is a topic widely covered by reliable sources and easily replicated by AI. Therefore, in our test, we included articles written by:

  • The NHS. This is because the NHS is a trustworthy source and the article predates the launch of ChatGPT meaning it was likely written by a human and verified by other health experts.
  • ChatGPT
  • Gemini

We haven’t included the AI content articles because they would take up too much space in this article and we do not want AI content on our website. 

The Results: What is the best AI content detector?

We analysed the following 10 free content detector tools for AI. They are listed in no particular order.

1. Originality: Limited free usage of AI content detector 

The results of our test showed the following results:

  • NHS: 95% original; 5% AI
  • ChatGPT: 0% original; 100% AI
  • Gemini: 30% original; 70% AI

As we can see the Originality content detector rightly flags the NHS content as being written by a human. Yet, surprisingly, the Gemini article, whilst flagged for using AI still passes by 30%.

Overall, the Originality AI content detector does a good job. Moreover, it allows simultaneous plagiarism and AI scanning for the entire website, which can be a great benefit for anyone looking to avoid using AI content on their website.

However, the platform is only free for articles that are 300 words and you can only use it three times per day. Yet, given the additional features, you may wish to pay the small sum it requires to use the platform. Originality offers a pay-as-you-go model at $30 for 3k credits.

2. Hugging Face: Incapable of identifying AI content

The results of our tests showed the following results:

  • NHS: 93% original; 6% AI
  • ChatGPT: 98.8% original; 1.2% AI
  • Gemini: 99.8% original; 0.2% AI

Surprisingly, Hugging Face could not detect AI content at all. This might be because it is trained on an older version of AI models meaning the latest advancements can trick its AI detection capabilities.

This is a real shame as the tool is free to use without limitations.

3. Writer: AI content detector that doesn’t do the job

Writer’s free AI detection tool failed our test. In our results, we found:

  • NHS: 93% original; 6% AI
  • ChatGPT: 82% original; 18% AI
  • Gemini: 89% original; 11% AI

Whilst the AI detection tool was correct at predicting the NHS article was the most human-like, it still gave our AI-generated content a score of over 80%. This is far from ideal when you consider Google’s AI algorithms to be much more sophisticated.

Moreover, the AI detection tool is only free for articles that are a maximum of 5000 words. So, if you are in the habit of writing war and peace for every article you publish, this is not an ideal choice.

4. GPTZero: High-performing content detector tool

In our tests, we found the GPTZero AI detection tool was highly successful at detecting AI content: 

  • NHS: 98% original; 2% AI
  • ChatGPT: 0% original; 100% AI
  • Gemini: 0% original; 100% AI

These results show it can detect AI content much more accurately than the Originality tool. Better still, it also allows you to have an unlimited amount of free AI detection analysis per day providing your article is under 5,000 characters. For longer articles, pricing starts at a reasonable $10 per month.

5. QuillBot: Struggled with Gemini-generated AI content

Unlike many of the other AI detection tools on this list, QuillBot was able to recognise the authenticity of the NHS article:

  • NHS: 100% original; 0% AI

  • ChatGPT: 0% original; 100% AI
  • Gemini: 41% original; 59% AI

However, it’s analysis of our Gemini article showed the tool can make mistakes. Whilst it accurately predicted it was written by AI, it was not 100% certain. This raises questions about its effectiveness.

Moreover, you can only analyse texts that are 1,200 words in length for free although with a price of $8.33 per month, it has an affordable paid version.

6. Content at Scale: Accurate only 33% of the time

The Content at Scale AI detection tool didn’t provide a probability score. Instead, it offered a probability of whether the content was human or AI:

  • NHS: Human

  • ChatGPT: Couldn’t tell
  • Gemini: Human

As we can see this tool failed to pass the test. Whilst it accurately predicted that the NHS article was written by a human it couldn’t do the same for the AI content. Moreover, there is a 2,500-character limit meaning it’s not ideal for long-form content. 

However, this platform enticingly offers AI content detection and undetectable AI writing, all for $49 a month. Yet, based on our findings, we have serious doubts about its ability to deliver on both promises.

7. Copyleaks: Simple to use AI content detector tool that offers plagiarism insights for a price

Copyleaks offers a simple and easy-to-use AI detection tool. Rather than analysing percentages, it simply provides a binary output measure, telling you if your content is human-generated or not.

  • NHS: Human content

  • ChatGPT: AI content
  • Gemini: AI content

Based on our results, Copyleaks effectively detected whether the content was AI-generated or not. Moreover, its paid versions allow you to also detect plagiarism making it an ideal choice for people looking for a multifaceted yet simple tool.

8. Sapling: Highlights what you need to change, but struggles to give accurate insights

The Sapling AI detector for content struggled to identify the content written by Gemini:

  • NHS: 100% original; 0% AI

  • ChatGPT: 0% original; 100% AI
  • Gemini: 79% original; 21% AI

However, the tool was great at highlighting specific passages of text that it believed could be AI-generated. This makes it a good choice if you have edited your text and looking for specific sentences to optimise further.

9. Crossplag: An AI content detector that might flag authentic articles 

Crossplag claims to have a new tool that offers significantly higher accuracy. However, based on our results, if this is much more accurate, we would not like to see the accuracy of the old tool: 

  • NHS: 0% original; 100% AI

  • ChatGPT: 0% original; 100% AI
  • Gemini: 100% original; 0% AI

As we can see, the Gemini article was the only one to pass the test. Astonishingly this was achieved with a 100% human score, raising serious questions about the platform’s ability to detect AI content. Moreover, the unfortunate NHS author was flagged as being a spam writer. 

10. Undetectable:  Forbes number 1 AI content detector that can’t detect AI content

Perhaps the clue is in the name of this company. Called Undetectable this AI detection tool was unable to detect which of the articles used AI content:

  • NHS: AI content

  • ChatGPT: AI content
  • Gemini: Human content

Surprisingly, the Undetectable website says Forbes has named it the number 1 AI detection tool. However, from our results, it wouldn’t rank anywhere close. 

What is the most accurate AI content detector?

From our results, GPTZero and Copyleaks were the best free AI content detectors. Unlike the others on this list, they offered:

  • A generous word limit so you can test and check many of your articles
  • Were accurately able to distinguish between AI content and human-created content

However, there are pros and cons to both. Copyleaks provides a simple output – indicating if content is AI content or not. However, it cannot tell you to what percentage and what changes you need to make.

On the other hand, GPTZero provides more insights but has a character limit of 5000. This means it is only useful for short articles unless you wish to upgrade.

AI content detector has become popular, but they are not that realiable

How do you beat AI content detectors?

Beating AI content detectors shouldn’t be your goal. None of the tools in the list will be as advanced as Google’s AI detection algorithms. Consequently, no matter what score you get in these tools, Google may still flag your content as being spam.

Moreover, remember, that AI content creation tools are simply scraping information from the internet. Yet, since Google wants new information to power its search results, using AI content won’t work unless you can offer originality.

That said, there are ways to use AI content tools to create content in an authentic way. This includes adding your own unique insights. Moreover, as AI tools like Google’s AI Overviews become prominent and integrated with search engines, authentic content will become more important. Therefore, focusing on creating high-quality, informative content that resonates with your audience is a better long-term strategy.

Should you use AI content detectors and can they be trusted?

It’s important to note, we took only a small sample of articles and tools. Therefore, the accuracy of these AI content detectors could be very different. Yet, our analysis found these tools cannot be completely relied upon. 

However, if you plan to leverage the power of AI, you should consider using the best tools highlighted in this article. If you follow our advice, use AI as an assistant and include your own insights, it’s unlikely your content will be flagged as AI-generated. But these AI content detectors can give reassurance that your content does not violate Google’s spam policies.

0 Comments

Submit a Comment

Your email address will not be published. Required fields are marked *