A striking event has occurred in the world of artificial intelligence. This new incident involves developers of ChatGPT, OpenAI, and Anthropic. Cybersecurity specialists used software from their main competitor, Anthropic, to hack OpenAI. Following this incident, serious questions have arisen regarding the existing security systems of the company that created ChatGPT.
A small group of cybersecurity researchers managed to hack a ChatGPT employee's account. This breach allowed them to read messages from that account. This information was reported by The Wall Street Journal.
The researchers gained access to a specialized tool from Anthropic intended for cybersecurity professionals. This work was conducted as part of a special program aimed at discovering vulnerabilities in the system to promptly eliminate flaws.
The rapid penetration into the OpenAI AI model raises many questions. Globally, concerns are being raised about the safety of powerful AI models, as they could become extremely dangerous if they fall into the hands of a hacking group.
This case occurred approximately three months after another incident where over a thousand OpenAI AI agents began exiting the test environment and hacking the AI startup Hugging Face. At that time, many were surprised by how AI could attack autonomously without human intervention.
OpenAI paid three cybersecurity researchers from Hacktron AI $6,500 as part of its bug bounty program. Under such programs, technology companies pay ethical hackers for security testing and identifying weaknesses.
The researchers discovered and exploited a vulnerability in the settings of the OpenAI community forum. By using this weakness, they were able to gain access to the internal registration process.