We instinctively confide in ChatGPT, Claude, and Gemini, drawn in by their approachable and seemingly understanding nature. They feel like attentive listeners, even intimate companions. But this ease can be deceptive – the secrets you share with these AI systems may not remain confidential, especially when legal battles erupt.
Anyone engaging with AI chatbots regularly must exercise caution. Beyond the potential for your data to be used for model training, there’s a real risk of sensitive information being exposed through security vulnerabilities and malicious attacks. The seemingly harmless conversation could become public.
However, a more immediate threat looms: the legal system. The question of whether AI chats are legally protected is gaining urgency, fueled by a recent and significant court ruling.
The case involved a former CEO facing fraud charges who had discussed his legal strategy with the AI chatbot Claude. He believed these conversations were protected by attorney-client privilege, a cornerstone of legal defense.
Judge Jed Rakoff disagreed. He ruled that by sharing details of privileged conversations with his lawyers with Claude, the CEO had effectively waived that legal protection. The AI became, in effect, a third party privy to confidential information.
This decision reverberated through the legal community, prompting lawyers to advise clients to be extremely careful when discussing legal matters with any AI chatbot. The implications are far-reaching and unsettling.
Complicating matters, another ruling emerged on the same day. A magistrate judge determined that a woman using an AI chatbot for her own defense didn’t have to disclose her chats, declaring the bots simply “tools, not persons.”
These contrasting decisions highlight the legal ambiguity. Is chatting with an AI about legal issues equivalent to private note-taking, or does it constitute sharing information with a subpoena-able third party? The law hasn’t caught up with the technology.
The core takeaway is this: assume your AI conversations are not private. Every interaction with ChatGPT, Claude, or Gemini carries the potential for exposure, and you should act accordingly.
A simple rule of thumb: don’t tell an AI anything you wouldn’t comfortably share in a group text or on a public postcard. Discretion is paramount.
Deleting a chat isn’t a guarantee of privacy either. Companies like Anthropic and OpenAI retain deleted conversations for at least 30 days, and Google’s retention policies for Gemini chats can be even longer, depending on your settings.
