ChatGPT data leaks

What Happens When Client Data Leaks Through ChatGPT?

Updated November 16, 2026

Short answer: Usually nothing dramatic happens straight away, which is why it goes unnoticed. ChatGPT privacy depends on the plan: the data sits in someone's chat history on OpenAI's servers, kept for as long as that account's settings allow, and on personal plans it may be used for training unless the user switched that off. The public incidents show how it then gets out: internal Samsung data was leaked to ChatGPT in 2023, a ChatGPT bug that year showed some users other people's chat titles, and in 2025 shared ChatGPT conversations turned up in Google search. For a firm, the harder ChatGPT data privacy problem is that you often can't get the data back or delete it, and you may have duties to clients or regulators once you know.

Samsung, 2023: internal data leaked to ChatGPT

In spring 2023, "internal, sensitive data from Samsung was accidentally leaked to ChatGPT." Samsung responded by "temporarily restricting the use of generative AI tools on company-owned devices," and asked staff using AI elsewhere "not to submit any company-related information or personal data." One of the problems Samsung noted was that it is difficult to "retrieve and delete" data on external servers (TechCrunch, May 2, 2023).

The lesson for a small firm: no hack was needed. The data went into a public tool by accident, and once it was in, the company could not easily pull it back.

OpenAI, March 2023: a bug showed other users' chat titles

OpenAI took ChatGPT offline after "a bug in an open-source library which allowed some users to see titles from another active user's chat history." OpenAI also found that the bug "may have caused the unintentional visibility of payment-related information of 1.2% of the ChatGPT Plus subscribers who were active during a specific nine-hour window" (OpenAI, March 24, 2023).

The lesson: anything an AI service keeps can be exposed by a mistake at the AI company, not only by a mistake in your office. The less a tool keeps, the less a bug like this can show.

Shared chats in Google search, 2025

ChatGPT had an option to make a shared conversation discoverable by search engines. After people found shared chats in search results, OpenAI removed it "to prevent users from unwittingly exposing sensitive information." OpenAI's security chief said: "Ultimately we think this feature introduced too many opportunities for folks to accidentally share things they didn't intend to, so we're removing the option" (The Register, August 1, 2025).

The lesson: sharing features are easy to misread. A staff member who shares a chat to show a colleague may not realize who else can open the link.

ChatGPT data privacy: what happens to data a staff member pastes in

It depends on the plan, which is why the first question after any slip is "which account?"

Personal ChatGPT (Free, Plus, Pro) ChatGPT Business
Used for training? "data sharing is enabled by default" until the user turns off "Improve the model for everyone" "By default, we do not use your business data for training our models."
How long it is kept Kept in chat history until deleted; temporary chats "may be retained for up to 30 days for safety purposes" Admins "can control how long your data is retained"; deleted chats removed "within 30 days", with legal and safety exceptions
Who can delete it Only the staff member who owns the account Your firm's workspace admins

Sources: OpenAI data controls FAQ and OpenAI enterprise privacy, read November 16, 2026. Claude's consumer plans work in a similar way: "If you allow us to use your chats or coding sessions to improve Claude, we may retain your data in a de-identified format for up to 5 years in our model training pipelines" (Anthropic).

The key row is the last one. When client data goes into a staff member's personal account, your firm has no admin control over it. You are relying on that person to delete it, and on the AI company's own deletion timetable after that. For the wider question of whether ChatGPT is private at all, see Is ChatGPT private? on Zero Data Retention AI.

What to do if it has already happened

  1. Find out exactly what went in. Which client, which documents, which tool, which account, and when.
  2. Delete it from the account, including any shared links. Deleting a chat may not remove it from the company's systems straight away, so note the date.
  3. Turn off training on that account if it is a personal one and the setting was on.
  4. Check your duties. Depending on your sector, state and client contracts, you may have to assess whether this counts as a breach and tell someone. The sector risks page points to the rules for insurance agencies, advisers and tax practices. Ask your counsel.
  5. Fix the cause. Usually that means giving staff an approved tool and a short policy. See How do I stop employees pasting client data into ChatGPT?

About Faradex. Faradex is a private AI workspace for small and mid-sized firms that handle confidential client work. Each firm gets its own Workspace on dedicated infrastructure in the United States. Files and conversations are held in memory and are gone when the Workspace is reset. AI model: Currently Opus via AWS Bedrock, more information is available on our website. AWS does not store prompts or responses, and model providers have no access to them. Pricing: from $120 a month, no seat minimum; see faradex.ai/pricing.