The best and worst AI for your privacy, ranked - and how each handles your data
Incogni researchers analyze 13 AI platforms and the potential risks they each pose to your privacy. The larger the platform, the bigger the risk - with one exception.
ZDNET's new study ranks 13 AI platforms based on their privacy risk, revealing potential concerns for user data. The largest AI players tend to be the most invasive, except for OpenAI's ChatGPT. Here's how each platform scored:
1. Vibe (formerly Le Chat, now Mistral AI): Offers easy-to-understand privacy policies and minimal third-party sharing, making it the safest option.
2. ChatGPT (OpenAI): Features a clear privacy policy, simple opting-out process, and transparent data usage for security and marketing purposes.
3. Pi (Inflection AI): Has a complex policy but handles EU users separately (GDPR), shares data with corporate and research partners, and maintains a privacy-friendly application.
4. Perplexity AI: Provides an easy-to-follow privacy policy, although it lacks detail and may share data with corporate affiliates.
5. Gemini (Google): Collects extensive data from other services, defaults to a vague privacy policy, and makes it difficult to find specific information about data handling.
6. Qwen (Alibaba): Describes data usage for analytics and search engines but is unclear about user data sharing and conversations' role in training.
7. DeepSeek: Allows local network storage for open-weight models, but privacy concerns increase with hosted versions. Users can request corrections for inaccurate personal information.
8. Z.ai: Offers open-weight models, but privacy details are vague, and hosted versions raise privacy concerns.
9. Gemini (Google): Collects vast amounts of data from other services, defaults to a general privacy policy, and makes it challenging to address specific questions about data handling.
10. Claude (Anthropic): Features an extensive privacy policy and data resources, with default opt-out for training; however, it may publish research using aggregated, de-identified user data.
11. Grok (xAI): Shares user data with corporate group members and affiliated parties, including conversations from public X platforms.
12. Kimi (Moonshot AI): Ranked as the greatest privacy risk among the assessed platforms due to unclear data handling practices.
Written by urgent.news from ZDNet's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
