OpenAI has announced the launch of a bug bounty program. The program encourages security researchers, technology enthusiasts, and ethical hackers to identify and report vulnerabilities, bugs, and security flaws in OpenAI systems.
As part of its commitment to developing secure and advanced AI, OpenAI is partnering with Bugcrowd, a crowdsourced security platform, to enable the bug bounty program. Interested parties can participate by registering with their email or using the “@bugcrowdninja.com” address.
OpenAI accepts or rejects 75% of applications within two hours, and the company awards a reward based on the severity of the problem. The bug bounty program rewards individuals from $200 for low-severity detections up to $6,500, and the maximum possible award for exceptional discoveries is $20,000.
However, the developers have listed several threats, the detection of which is not rewarded. The program’s official page says: “Issues related to the content of model prompts and responses are strictly out of scope and will not be rewarded unless they have an additional directly verifiable impact on security within the service.”
Some experts have pointed out that the initiative is unlikely to eliminate a wide range of cybersecurity risks. Rather, the program will help OpenAI reduce the number of internal security problems. Nonetheless, the bug bounty program is crucial in ensuring the safety, security, and trustworthiness of OpenAI’s services.