Similarly,
What ai best protects your:
In terms of IA chatbots. Consequently, a fundamental question remains: what is the real counterpart for our personal data? Nevertheless, The artistic vagueness (voluntary?) Privacy policies makes any comparison as hazardous as it is tedious. For example, But that did not prevent incogni (a data deletion service) from rolling up to offer us a study that screened new from the greatest AI systems. In addition, in order to assess their respect for privacy. Consequently, The verdict of this first edition does not overly surprise by placing a European actor in mind. Nevertheless, pinning the American jerseys.
Incogni claims to have applied a rigorous methodology, articulated around three main axes and eleven criteria. Meanwhile, Thus. For example, each platform has been assessed about its specific IA privacy policy (the most heavily weighted category), its transparency and its general data collection.
The discreet cat – What ai best protects your
Less known to the general public than its American competitors, it is the cat that seizes the first place in this ranking. However, The model of the Parisian company Mistral AI is therefore the most respectful chatbot of privacy according to this study. Meanwhile, thanks in particular to a much more sober approach in terms of data. For example, French AI only collects our personal information in a limited manner. Better still, it only shares user prompt with service providers that are essential for its proper functioning.
On the second step of the podium, we find the very popular Openai Chatgpt. It rises also high thanks to a major asset: exemplary transparency. It allows users to easily understand where their data go and how to limit their use. The nuance that differentiates what ai best protects your it from its French rival is due to the perimeter of sharing: where the cat sticks to its providers. Chatgpt can also communicate information to its “affiliates”. Behind them, we find Grok, Xai’s solution, and Claude, developed by Anthropic.
Bad students not so surprising
On the other hand, the report points to a very problematic practice at Meta, Microsoft and Google: the use of a unique privacy policy for all of their products. It then becomes very complicated for the user to find information specific to AI platforms since they are lost. in a legal jargon as dense as it is generalist.
The platforms designed by the largest American tech companies therefore prove to be the most invasive. In the last position, Meta AI was deemed the worst of nine AI evaluated. Meta’s service what ai best protects your is not only qualified as “eager for data”. but it also widens the gap in collection and sharing, displaying an almost double risk score of that of the penultimate, Gemini.
Just in front of Meta, Gemini from Google ranks eighth and Microsoft Copilot comes in seventh position. Their poor performance is explained by the volume of data they amass and the lack of clarity of their practices. And this does not only concern prompt. since Meta AI and Gemini’s mobile applications can for example collect information as sensitive as precise location, postal addresses or telephone numbers. For his part. Copilot obtains the worst note on the criteria purely linked to AI, in particular on the use of conversations for the training of models.
Can we prevent AI from learning from our conversations?
This is undoubtedly the question that many of you are posing when reading this article. what ai best protects your For the most concerned about their data, some platforms play the game. It is indeed possible to deactivate this option (an “opt-out”) on the cat, chatgpt, grok and co-pilot. These services then make it possible to converse with AI without these exchanges nourishing its future learning.
On the other hand. the situation is complicated in others since this possibility does not seem clearly offered by Gemini, Deepseek, Pi Ai and Meta Ai. For the latter, the prompts you submit are therefore likely to be reused, without a simple mechanism to oppose it. A special mention all the same for Anthropic, who claims to never use user entrances to form his models.
Ultimately. one wonders if such a study can move the lines in the worst students in the ranking if it is gaining momentum. We are quite pessimistic on the subject. The race for AI is fierce and what ai best protects your data is one of the nerves of war. It is not for nothing that Meta has accelerated in terms of data collection linked to the AI lately. even in Europe, even if it means taint even an even more blurred public image on the subject.
Further reading: Quebec rocks could be the oldest on earth – Photos | The oldest rocks of the land discovered in Quebec – IP68, XXL battery, Snapdragon 8S Gen 4…: The Poco F7 hits hard despite its low price – Diabetes: Small cactus, an application from here to simplify life – Ecoflow launches Rapid Pro, a high -capacity feed bank with integrated cable, and offers discounts for first buyers.
