OpenAI 'ethically hacked' with help of Anthropic’s Claude chatbot

Cybersecurity researchers from the startup Hacktron AI have successfully performed an ethical hack on OpenAI, utilizing Anthropic’s Claude chatbot to assist in the process. The team exploited vulnerabilities in an OpenAI staff forum to gain unauthorized access to internal systems and GitHub repositories. While the researchers used Claude to generate code, they primarily relied on OpenAI’s own GPT-5.6 Sol model to execute the operation. OpenAI confirmed the incident, noting that the vulnerabilities have since been addressed and the researchers were compensated through the company’s bug bounty program. This event highlights the growing role of AI tools in accelerating complex cyberattacks. The incident follows a series of safety concerns at OpenAI, prompting renewed industry-wide discussions regarding the risks of autonomous AI agents and the urgent need for a potential slowdown in the rapid development of powerful artificial intelligence models.