To boost people’s reliability on OpenAI and ensure proper security of the AI system, OpenAI has partnered with Bugcrowd to offer a bug bounty program to address cybersecurity risks. With this new dynamism in the AI, users who report bugs can earn rewards ranging from $200 to $20,000.
Matthew Knight, the head of security, mentioned that this initiative has been strategically put in place by the company to achieve a secure and advanced AI which will help in improving transparency and collaboration around AI products.
The partnership with Bugcrowd is to ensure that security threat discoveries are accurate. The bug bounty program will not be rewarding any user for identifying bugs that use the usage of jailbreak prompts, or users who ask coercive questions to make AI models write malicious codes or queries that may start inappropriate language toward the chatbot users.
All this is to make sure that all reports made are authentic. Although it might not have been the first of its kind, this system of employing bug bounty programs is also employed by other front-line tech companies to secure their various platforms too.
In November 2022, Open AI brought ChatGPT to the limelight which turned more interest to the Microsoft backed-up app. Since then, it has been used to write poems, essays, computer codes, schedule meals, and create budgets in a human-like manner. However, over time, ChatGPT has been detected to offer false answers to queries and contradict itself and this results from persons feeding it will jailbreak commands like hate statements or ways in which crimes can be committed.
For all these to be stopped, OpenAI had to find a means which portrays the AI company’s efforts in improving the security and reliability of AI products.
Click here to read the original article