Hacker Accesses OpenAI’s Internal Messaging System. Adding Fuel to Fire as Security Is Under Scrutiny

A hacker managed to access confidential information from OpenAI, specifically the private messaging between its employees, in early 2023, when the company had already put ChatGPT on the market, according to reports The New York Times.

Because it is importantThis incident reveals security vulnerabilities at one of the largest AI companies. It also raises questions about the possible theft of technological secrets by a major adversary in its field, such as China.

The contextThe hacker gained access to internal messaging forums where employees discussed the technologies they were developing. He was unable to access either the source code or the company’s core AI systems.

OpenAI discovered the incident but merely reported it internally. It did not make it public or share it with the FBI, understanding that only messaging had been accessed, not code.

Key concerns.Primarily, the possibility of China or other adversaries stealing advanced AI technology from a major U.S. company. Also questioned are the strength of OpenAI’s security measures, as well as the risks inherent in AI development and how to manage them.

Critical voices. Former OpenAI employee Leopold Aschenbrenner commented on the incident in a podcast. He also said the company was not doing enough to protect itself from foreign espionage. Then he was fired.

Industry experts such as Daniela AmodeiAnthropic’s team argues that the current risks of AI are not that dramatic, and that sharing knowledge can be beneficial for the entire industry.

The futureOpenAI began to strengthen its security following the incident and created a specialized committee, although it has been under scrutiny for months precisely because of its approach to security.

The question now remains whether stricter government regulations are necessary to prevent espionage in a field as sensitive as AI, especially with rival countries such as Russia or China in mind.

Featured image | Xataka with Midjourney

At Xataka | Security is one of the current major shortcomings of AI. The ChatGPT client for Mac is the latest example