ChatGPT Plugin Vulnerabilities Exposed Data, Accounts

ChatGPT Plugin Vulnerabilities Exposed Data, Accounts

March 13, 2024 at 09:27AM

Salt Security analyzed ChatGPT plugins and uncovered vulnerabilities that could be exploited to access sensitive data and take over accounts on third-party websites. These vulnerabilities affected the OAuth authentication process, potentially leading to unauthorized data access and account takeovers. Vendors were promptly notified and patches were implemented. Additionally, future GPTs are expected to replace plugins, but Salt Security also found vulnerabilities in GPTs.

Based on the meeting notes, it is evident that Salt Security conducted an analysis of ChatGPT plugins and identified several vulnerabilities that could have been exploited to obtain potentially sensitive data and take over accounts on third-party websites. These vulnerabilities included issues related to OAuth authentication, which allowed attackers to install malicious plugins with their own credentials on victims’ accounts and gain unauthorized access to user data.

The vulnerabilities were reported to the relevant vendors, including OpenAI, PluginLab.AI, and Kesem AI, and patches were subsequently rolled out. It’s worth noting that when Salt Security conducted its research, ChatGPT plugins were the primary means of adding functionality and features to the LLM, but OpenAI later announced the introduction of customizable GPTs for paying customers, which are expected to replace the plugins.

Furthermore, Salt Security has also identified vulnerabilities in GPTs and plans to outline these in an upcoming blog post. The notes also mention that other parties have found ways to abuse GPTs to obtain potentially valuable data.

In summary, the meeting notes highlight the critical vulnerabilities discovered in ChatGPT plugins, the actions taken by Salt Security and the vendors to address these vulnerabilities, and the shift towards customizable GPTs to replace the plugins in the future. Additionally, it is noted that Salt Security plans to release details about vulnerabilities found in GPTs and the potential abuse of GPTs by others.

Full Article