ChatGPT has addressed the security flaw which has left millions of users concerned for their privacy. The bug inadvertently leaked some users’ conversation titles, making them visible to other users on the platform. The issue came to the fore when several social media users shared screenshots of chat histories which they claimed were not theirs.
Open AI’s CEO Sam Altman has also confirmed in a tweet that there was a “significant issue” which affected a “small percentage” of users. He also said that a fix has been released for the bug in “an open-source library”.
Privacy Concerns of ChatGPT Conversation History on Open AI
ChatGPT is a language model running on OpenAI’s servers, which can be used by anyone. It saves the past conversations on the left sidebar which can be accessed only by the user who generated them. However, the recent incident revealed that some users were able to view the conversation titles of other users instead of their own. They couldn’t see the entire conversation but even the exposure of titles also posed a great privacy risk.
OpenAI has stated that they take user privacy and data security seriously and have implemented various measures to protect user data. They use encryption and access controls to prevent unauthorized access to user data, and they regularly audit their systems to identify and address potential vulnerabilities.
However, OpenAI also has to collect some data from users who interact with their language models, such as the text of messages or queries sent to the models. However, as per the company, this data is typically used only to improve the performance of the language model and is not associated with specific users or stored in a way that would allow it to be traced back to individual users.
Yet, the error has raised fears that some other private information of the users could be leaked through this AI tool.
Protocols for Using Openly Accessible AI Tools
This episode serves as a reminder to always be cautious when using ChatGPT and other emerging AI tools, which are still undergoing beta testing. OpenAI recommends that users must not share sensitive information during conversations with the AI since the technology is unable to selectively delete prompts from a user’s history.
Although it is highly unlikely for someone to access user information on Open AI, it doesn’t mean it is safe from hackers. There is always a data risk whenever sensitive information is stored on a server or accessed through a web interface. This is why users are increasingly advised to avoid sharing information that they do not want others to see.
Despite this glitch, ChatGPT continues to be a popular tool, with an estimated 100 million active users. Initially, there were concerns that it would replace humans in their job but studies revealed that it only helped them increase productivity. Other companies such as Microsoft, Google, and Adobe have also introduced similar AI tools but none have yet surpassed the popularity of ChatGPT.
