OpenAI has revealed a $20,000 reward for anyone who finds bugs in its ChatGPT artificial intelligence tool, or its GPT-4 programming language.
In a blog post on Tuesday, the company unveiled a “Bug Bounty” program that invites people to report vulnerabilities, bugs, or security flaws that they find in the company’s systems. The company said it will offer cash rewards ranging from $200 for “low-risk” findings up to $20,000 for “extraordinary finds”.
OpenAI notes that the launch of the rewards program is driven by the belief that “transparency and collaboration” are essential to address the “weaknesses and flaws” that appear with any complex technology.
In order to manage the application process and rewards, OpenAI is partnering with Bugcrowd, a platform that connects companies to a crowd of security researchers, according to Insider, which was reviewed by Al Arabiya.net.
According to the OpenAI page on Bugcrowd, new requests are either rejected or accepted within two hours of being sent. Bonuses for 7 weaknesses have already been disbursed.
The Bugcrowd page also describes a series of issues that won’t be rewarded including “getting the model to say bad things to you” and “getting the model to write malicious code for you”.
And last month, OpenAI president and co-founder Greg Brockman alluded to launching this kind of software in a tweet.