The function to delete correspondence in popular AI services, such as ChatGPT and Claude, is merely an interface solution and does not ensure the physical deletion of data from servers. According to materials from the New York Times v. OpenAI lawsuit, dialogue logs can be preserved and provided upon court order as legally significant business records.
What Happened
During the New York Times v. OpenAI proceedings, it was established that even if a user deletes a chat in the interface, the company can retain this data. The court ordered OpenAI to provide conversation logs that were previously considered deleted by users. This confirms that deletion mechanisms in current LLM services perform only a logical hiding of information from the user interface, rather than its irrevocable erasure from storage systems.
Context
Unlike interactions with doctors or lawyers, dialogues with artificial intelligence are not protected by professional secrecy laws. Legally, such correspondence is treated as ordinary business records. This creates a gap between user expectations of privacy and the actual legal capabilities of extracting data from the cloud storage of AI companies.
Why This Matters for the Industry
For the AI industry, this precedent changes privacy standards by turning user data into potential evidence. Developers will have to revise data retention policies and find a balance between privacy requirements and legal obligations. In the long term, this may stimulate demand for privacy-first solutions and local on-premise models, where data control is physically guaranteed.
Why This Matters for Users
It is important for users to understand that any sensitive information—from medical questions to corporate strategies—can be preserved and subpoenaed by a court. When using ChatGPT or Claude, 'deleting' a chat only hides it from your list but does not guarantee confidentiality before the law.
Sources
Author
Look at AI, Editorial Staff