Researchers bypass ChatGPT via cloud

Cybersecurity researchers managed to breach OpenAI's internal system using "Cloud," an AI program from Anthropic, as part of a bug bounty program. The breach resulted in access to an employee's account at OpenAI and internal source code, achieved at a cost of less than $3,000 and in a short amount of time. This highlights the growing threats posed by advanced AI models and raises concerns about the safety of this technology, despite the company awarding the researchers a prize of $6,500.

Size: 120%Text Size

Notice: This Is an AI-Generated Summary

Display The Full Article