AI Privacy Report Card Which Chatbots Pass
As artificial intelligence becomes a daily tool for millions, concerns about data privacy are growing louder. What really happens to the questions you ask and the information you share with chatbots like ChatGPT or Gemini? A revealing new study from the research team at Incogni has analyzed the privacy practices of major AI platforms to see which ones protect users best—and which ones are the most invasive.
The study evaluated AI chatbots on several critical factors: how they use your prompts, whether they share your data with third parties, and how transparent their privacy policies are.
The Ultimate AI Privacy Showdown Best to Worst
The overall results show a clear gap between dedicated AI companies and the tech giants. French AI platform Le Chat (from Mistral) earned the top spot for being the least privacy-invasive. It was followed closely by ChatGPT and Grok, which were praised for transparency and user control.
At the bottom of the list are the AI tools from big tech. Meta AI was ranked as the worst for privacy, with Google's Gemini, Microsoft's Copilot, and DeepSeek also performing poorly.
Is Your Data Training the AI
One of the biggest concerns is whether your conversations are used to train future AI models. According to the study, Gemini, DeepSeek, Pi AI, and Meta AI do not allow users to opt out of having their prompts used for model training. Once your data is in a training set, it's nearly impossible to remove.
On the other hand, ChatGPT is the most transparent about this process and offers a clear way to opt out, a feature also available on Copilot, Mistral AI, and Grok.
Who Else Sees Your AI Data
The research also examined how AI firms share your data. It found that information is often provided to a wide range of third parties, including service providers, government bodies, and other businesses. Here again, Meta AI and Microsoft’s Copilot stand out for sharing data with multiple parties, expanding the circle of who might have access to your information.
Navigating the Fine Print of AI Policies
Transparency is key to trust, but not all privacy policies are created equal. The study found OpenAI's ChatGPT to be the most transparent, with easy-to-understand information. In contrast, larger companies like Google, Microsoft, and Meta often bury their AI policies within broader company-wide documents, making it cumbersome for users to find and comprehend the specific terms for their AI tools.
Beyond Prompts What Else Are AI Apps Collecting
It's not just your chats that AI companies are interested in. The mobile apps for these platforms collect a variety of other data. Pi AI and ChatGPT were found to be the best, collecting very little information beyond your conversations.
However, Gemini and Meta AI collect precise location and addresses. Meta AI also gathers usernames, email addresses, and phone numbers, while Gemini, Pi, and DeepSeek also want your phone number. Grok's Android app can access user-shared photos and app interactions.
Given these findings, it's crucial to review the privacy settings for any AI apps you use. For a deeper dive, you can explore the full Incogni study at this link.