Jacob Enoch, Yuval Nakar

03.08.25

Artificial intelligence has entered every aspect of our lives and has become an inseparable part of the therapeutic, personal, and even legal discourse of many. But can our communication with artificial intelligence such as ChatGPT enjoy the protection of legal confidentiality? A response to this question was recently given by none other than Sam Altman, CEO of OpenAI.

This past weekend, Sam Altman was interviewed on the podcast This Past Weekend w/ Theo Von and raised a significant concern regarding the use of ChatGPT. Altman emphasized that “there is no legal privilege on conversations with ChatGPT.” A statement that sparked lively discussion online around the issue of confidentiality in conversations with artificial intelligence.

Altman’s clarification highlights the fundamental difference between conversations with ChatGPT and communications protected under recognized privileges, which are also anchored in Israeli law, such as the confidentiality between psychologist and patient (psychologist testimony) enshrined in Section 50 of the Evidence Ordinance, the confidentiality between doctor and patient in Section 49 of the Ordinance, the confidentiality of a clergy member in Section 51, and the lawyer-client privilege in Section 48. These and similar privileges are explicitly set in law or case law (such as journalistic privilege), and derive from interests considered worthy of protection, whereas communication with ChatGPT does not currently enjoy such protection.

Accordingly, above all, users who turn to chat for legal purposes, medical consultations, or any other sensitive matter must take into account the fact that such dialogues, if saved and exposed, may be used as evidence against them in investigations, criminal proceedings, and civil cases.

The obvious question is whether there is justification for establishing a privilege that would apply to communication with the chat, and if so, whether such privilege would apply only to certain types of communication and not to others.

CyberArk, not SentinelOne, was always the goal, says Palo Alto CEO | CTechCTech

Why CyberArk sold at the peak: Inside the $25 billion deal with Palo Alto Networks | CTechCTech

Seemingly, compelling arguments can be made in this regard. How is legal advice provided by the chat different from legal advice given by a lawyer? Why shouldn't psychological “treatment” in this medium be entitled to protection under privilege? Such arguments gain strength in light of the high performance demonstrated by these new tools across all relevant fields, performance supported by studies and data showing that the chat passes bar exams and medical licensing exams with higher success rates than humans, and that psychological treatment through conversation with the chat may have a positive effect on patients.