Specialists from Hacktron AI were able to gain access to an OpenAI employee’s account in ChatGPT and the company’s internal repository on GitHub by using an Anthropic Claude neural network for this. This is reported by The Wall Street Journal.
The hack became possible due to a vulnerability in the Discourse service—through it, researchers obtained authorization tokens that turned out to be usable to sign in to OpenAI accounts. This involves white-hat hackers: after discovering a loophole, they did not exploit it to harm the company, but instead passed the information to OpenAI in exchange for a reward.
The hack became possible due to a vulnerability in the Discourse service—through it, researchers obtained authorization tokens that turned out to be usable to sign in to OpenAI accounts. This involves white-hat hackers: after discovering a loophole, they did not exploit it to harm the company, but instead passed the information to OpenAI in exchange for a reward.