The recent news of Microsoft and OpenAI catching state-sponsored hackers has reawakened concerns among ChatGPT users regarding how much the company spies on their chats.
Last week, Microsoft, which is the largest investor in OpenAI with a 49% stake, released a report saying they caught Chinese, Russian, and Iranian hackers using LLMs to augment cyber operations.
“In collaboration with OpenAI, we are sharing threat intelligence showing detected state-affiliated adversaries—tracked as Forest Blizzard, Emerald Sleet, Crimson Sandstorm, Charcoal Typhoon, and Salmon Typhoon— using LLMs to augment cyberoperations,” Microsoft said.
According to the report, no “identified significant attacks” were made using the LLMs, and all associated accounts have been blocked. While positive, however, it would not have been possible without the companies analysing chat records of the accounts involved.
Can OpenAI Spy on ChatGPT Users?
So, does it mean that OpenAI can access and read your conversations on ChatGPT? Well, yes. ChatGPT’s terms of service permit the developers to store and review users’ chats for different reasons. However, the extent to which it does it remains unknown.
“As part of our commitment to safe and responsible AI, we review conversations to improve our systems and to ensure the content complies with our policies and safety requirements,” according to ChatGPT’s FAQ page.
OpenAI may review ChatGPT conversations for two major reasons. Firstly, the developers store and review users’ conversations to train and improve future models. OpenAI said the data are anonymized, and the personal information of users in conversations is reportedly removed.
Secondly, OpenAI reviews ChatGPT conversations to disrupt illicit usages, as in the case of Chinese, Russian, and Iranian hackers.
Should Users Be Worried?
OpenAI’s data collection isn’t entirely a thing of concern, especially for an average user who frequents the tool for questions, research, and other basic activities. However, this equally comes as a reminder not to disclose sensitive information to the chatbot, as chats can possibly be reviewed.
It’s also worth noting that OpenAI has made provision for ChatGPT users to remove their conversations from OpenAI’s training data entirely.
