Home / Technology / AI Platforms: Big Names, Big Privacy Risks?
AI Platforms: Big Names, Big Privacy Risks?
20 Aug
Summary
- Incogni's study ranks 13 AI platforms on privacy risks.
- Larger AI platforms generally pose higher privacy threats.
- Users should treat all AI inputs as shared with third parties.
New research from Incogni, a data broker removal service, evaluates the privacy risks associated with 13 generative AI and large language model (LLM) platforms. The study, titled "Gen AI and LLM Data Privacy Ranking 2026," assigns scores based on privacy-invasive practices, data-sharing transparency, and data handling.
The findings suggest that major AI platforms generally present greater privacy concerns, with Gemini and Meta AI receiving higher risk scores. Conversely, Vibe, ChatGPT, and Pi demonstrated lower risk. However, no platform currently offers users the ability to retrieve their data once used for training LLMs, and retroactive consent withdrawal is not possible.
Incogni's information security manager Miguel Fornés advises users to consider all inputs to LLMs as third-party information. He recommends avoiding the sharing of sensitive data like passwords, financial details, or confidential work information. Fornés also suggests paying close attention to providers' privacy practices and, where feasible, opting for locally run models that keep data on-device.
