OpenAI's ChatGPT chatbot had security flaws storing conversations locally, prompting encryption update.

From CXO Media: 2024-07-08 02:12:54

OpenAI’s ChatGPT chatbot, made available to all macOS users, was found storing user conversations locally in plain, unencrypted text. This security flaw was highlighted by a Swift developer, prompting OpenAI to release an update adding encryption to stored chats to mitigate potential malware risks and vulnerabilities.

A hacker gained illicit access to OpenAI’s internal messaging systems in 2023, stealing AI tech designs and internal discussions but not AI build data. Concerns were raised about security vulnerabilities and possible exposure to foreign adversaries.

Former OpenAI technical program manager Leopold Aschenbrenner was fired for leaking information and raising concerns about security vulnerabilities. OpenAI maintains that his separation was not due to his concerns, and they dispute his claims regarding the company’s security.

Questions were raised about Big Tech companies’ ties to China following testimony from Microsoft President Brad Smith about Chinese hackers targeting US Federal Government networks. Tightened security measures are necessary to protect data, while restrictions on policy researchers from other countries could slow AI progress.

OpenAI acknowledges the need for top talent to advance AI technology but recognizes the associated risks. As the debate continues on security measures and the pace of AI advancements, the world must stay vigilant while holding our collective breaths.



Read more at CXO Media: OpenAI and Its Gaping Security Issues