Experts Warn Users to Limit Sensitive Data Shared with AI Chatbots Like ChatGPT and Google Gemini

Recent investigations into AI usage, highlighted by the Meli and Liel Yehulomi case, have raised widespread concerns about what happens to personal data shared with AI chatbots such as ChatGPT and Google’s Gemini. Authorities reportedly traced conversations and activities conducted via AI systems, including inquiries about anonymous phone purchases and apps that could compromise privacy. Although some chats were deleted, investigators managed to recover parts of the data, prompting questions about the permanence and privacy of AI interactions.

Cybersecurity expert Ofer Cohen, founder of YO SECURE, explains that deleting a conversation in AI platforms does not guarantee immediate or complete erasure. For example, OpenAI states that deleted ChatGPT conversations are removed from user views instantly but may remain on company servers for up to 30 days or longer for legal or security reasons. Google Gemini’s data retention depends on user account settings, allowing users to manage deletion periods. Therefore, deletion is important but does not imply total disappearance from all systems.

Regarding privacy, Cohen emphasizes that while AI conversations are not routinely monitored by humans, users should not treat these chats as entirely private diaries. AI services operate through accounts, devices, and servers governed by terms of use and privacy policies, meaning shared information is stored digitally and potentially accessible under certain conditions. Unlike traditional search engines, AI chats often contain detailed personal reflections, documents, and sensitive information, making the data more revealing.

Law enforcement access to AI chat data requires proper legal procedures; not all conversations automatically become evidence or are freely accessible. Digital investigations often rely on a combination of technical data such as IP addresses, login times, and backups rather than single messages. Cohen advises users to treat AI tools like cloud services and avoid submitting passwords, credit card details, or confidential business and personal information unless absolutely necessary.

The expert also highlights workplace risks, where employees might upload sensitive company data to AI tools, potentially exposing it externally. Organizations should establish clear policies on AI usage, data sharing limits, and account management. As AI becomes integral to daily life, users must mature their approach to data privacy by consciously limiting what they disclose to AI systems. The key takeaway is to ask whether the AI truly needs specific personal details to assist, and if not, to withhold them to protect privacy in the AI era.