AI LOGIK, a French startup, recently introduced WAIZZ AI Detector, a tool designed to identify texts produced by artificial intelligence. This announcement comes at a pivotal moment: the massive rise of generative models like ChatGPT, Gemini, and Grok 3 has made distinguishing between human and artificial content increasingly challenging, especially in fields where information reliability is paramount. Education, journalism, and businesses are directly affected.

WAIZZ AI Detector: Claimed 99% Accuracy

The team behind WAIZZ AI detector asserts an impressive effectiveness, claiming a “99%” accuracy rate after comparative analysis of over 2,000 samples from major existing language models (LLMs). Compared to existing solutions like Quillbot or Originality AI, this detector stands out with its novel algorithm combining eight different metrics, aiming to minimize two common types of errors: false positives (when human-written text is mistakenly identified as AI-generated) and false negatives (the reverse case).

AI LOGIK specifies that, among the tests conducted, 99% of texts from large language models were correctly identified, a performance that, if confirmed, would position WAIZZ strongly in this sector.

A Multilingual Detector

A distinctive feature of WAIZZ AI Detector is its handling of six languages: French, English, Spanish, Italian, German, and Portuguese. While most similar tools remain focused on English (with some limited extensions), WAIZZ seeks to cover less addressed languages like German or Italian, areas where other solutions often struggle to maintain their reliability. This adaptation to the grammatical and stylistic specifics of each language is a considerable advantage. The benefit is concrete: a teacher in France can easily examine a paper written in Portuguese, and conversely, a German editor can analyze a suspicious text without the usual limitations imposed by some exclusively English-language detectors. This is a clear need in the current context of increasingly international and automated text production.

The approach adopted here is based on a clear desire: to make the tool accessible to as many people as possible while removing immediate financial constraints, at least in its initial key functionalities. Thus, anyone can analyze up to 2,000 words without even creating an account. By registering for free on the WAIZZ AI Detector platform, this threshold then increases to 5,000 words per request, with still no imposed limit on the total number of analyses possible per day or week.

This strategy contrasts sharply with those favored by some competitors like Copyleaks, where the rapid shift to a paid offer drastically restricts available free trials. According to Fabrice COQ, CEO of AI LOGIK, offering a solid free base would not only allow users to better understand the detector’s effectiveness but also promote its gradual adoption within the vast ecosystem offered around WAIZZ, which is not limited to detection but also integrates various tools related to the generation of all types of automated content.

How Do Existing Solutions Measure Up?

Recent comparisons conducted on various AI detectors show that even among current references, including Quillbot Detector or ZeroGPT, performance rates generally oscillate between “78%” and “84%”, according to several studies published in early 2024. These are decent performances but subject to notable limitations as soon as the field extends beyond strictly standard English or when it comes to accurately evaluating texts written by non-native English speakers.(AI-Detectors Biased Against Non-Native English Writers | Stanford HAI)

WAIZZ AI Detector thus claims a higher margin with its announced “99%”. However, these figures still require external validation through independent studies to be fully credible.

WAIZZ: An AI Gem to Watch

By integrating a technical method that combines innovation and simplicity, in a context where trust remains a priority, AI LOGIK positions WAIZZ in an approach that goes beyond detecting whether a text has been artificially generated, but also provides a more nuanced measure: a “humanity score”, intended to accurately estimate the likelihood that content is human-generated.

The interest in this type of indicator seems evident in an environment saturated with automated texts. Being able to gauge the probable degree of artificial intelligence behind each production could become, for certain actors (journalists, teachers, businesses), an indispensable resource.

With this launch, AI LOGIK not only confirms its ability to develop high-level AI tools but also attempts to take a strategic position in the AI detection market and, more broadly, in generative AI.