UncovAI introduces an innovative platform for detecting Generative AI content, offering a blend of efficiency and sustainability. Instead of relying on resource-intensive Deep Learning models, we employ mathematical hypotheses that effectively distinguish between AI-generated and human-authored content.

Today, the challenge is to propose generative models capable of reproducing human content as closely as possible. To achieve this, major companies are investing colossal sums of money and have made successful models available like GPT-4, Claude … According to one study, 79% of employees are familiar with generative models, and 22% use them daily. According to another study, by 2025, 10% of all data will have been generated using generative models. 

Therefore, It will be crucial to distinguish the origin of the data, human-authored or AI-generated. UncovAI is already offering a sustainable solution to this emerging question.

These days, we are beginning to see the impact of global warming and have a better understanding of its causes. New models, particularly Deep Learning models, have a significant energy impact. Hence, employing power-hungry models for AI content detection (generated with power-hungry models), considering their energy consumption, is inconceivable.

That’s why UncovAI has developed an efficient detection model that has the smallest carbon footprint. To fulfil one of the most significant missions of all: to help everyone find the origin of content while respecting the environment.

As reported by our users, UncovAI can be valuable in many use cases, such as education, recruitment, social networks, journalism and beyond. 

As soon as you engage in a conversation with someone or ask for work or homework and you don’t feel that the content was human-written, you can use UncovAI.

UncovAI works robustly across a range of GenAI models, including but not limited to GPT-like models (ChatGP, GPT4, GPT3, …), LLaMA, Claude and AI services based on those models for text and  DALL-E, Midjourney, Leonardo, Ideogram or Canva, and other AI models for image.

UncovAI works well for English and French. Our first studies show that our model is working with the Ukrainian language.

Try with your language! Let us know how it worked!

UncovAI solution for detecting GenAI texts will give five types of answers:

• 🤖 Most likely generated by a GenAI
• 🤖 Likely generated by a GenAI
• 😐 Difficult to determine
• 🧠 Likely written by a human
• 🧠 Most likely written by a human

UncovAI solution for detecting GenAI images will give two types of answers:

• Likely Generated
• Likely Human

Depending on the answer, UncovAI will have determined who is more likely to have made the content.

We would like to draw your attention to the fact that, as generative models evolve rapidly, UncovAI results should not be taken at face value but should provide food for thought. Moreover, We noted that pictures taken with smartphones or other tools that use AI processing can be classified as Likely Generated.

No, your content will not be stored. At UncovAI, we prioritize user privacy, and we want to assure you that your submissions are treated with confidentiality and will not be retained or used beyond the scope of the specific interaction. Your content remains your own, and our production model is designed to respect your privacy and data.