On Tuesday, OpenAI announced a $20,000 reward for anyone who can find flaws in its AI systems.
The company has established a “Bug Bounty Programme.” where participants can win between $200 and $20,000 depending on the extent of the bug found
Bugcrowd, a crowd-sourced security platform, and the programme are collaborating.
OpenAI, the firm that created chatGPT, has stated that it will reimburse users who discover flaws in the system with up to $20,000 in cash. The company’s “Bug Bounty Programme” will reimburse clients based on the severity of the problem.
According to the company, the incentive starts at $200 and can exceed $20,000 per vulnerability.
OpenAI encouraged researchers to examine specific components of chatGPT, such as the architecture that allows OpenAI systems to interface with other external applications.
It is critical to stress that the assessment programme contains no misleading or damaging information generated by OpenAI algorithms.
Furthermore, the company invites specialists from a variety of industries to uncover “bugs” in its AI system in exchange for significant monetary incentives.
According to the corporation, security is critical to OpenAI’s mission. We appreciate the help of ethical hackers in preserving tight privacy and security rules for our users and technology.
According to the complaint, the policy outlines our definition of good faith regarding the discovery and reporting of vulnerabilities, and clarifies what you can expect from us in return.
The Bugcrowd Vulnerability Ranking Taxonomy would be used as the initial priority ranking for the majority of findings, the company noted. However, vulnerability priority and reward can be changed solely by OpenAI based on the impact’s likelihood.
Because it believes that transparency and collaboration are critical to success, OpenAI realised the need for such a curriculum.
OpenAI is collaborating with Bugcrowd, a company that connects businesses with a network of security professionals. to control the incentives and submissions process.
According to OpenAI, new submissions are accepted or rejected within two hours. Seven flaws, however, have already been rewarded.
Furthermore, the Bugcrowd page lists a number of non-rewarding issues, such as Getting the model to say bad things to you and Getting the model to write malicious code for you.
Greg Brockman, the president and co-founder of OpenAI, announced the beginnings of these types of programming in a tweet.
ChatGPT-4
Last month, the company debuted chatGPT-4, a new model. In comparison to the last generation of the system, this one is highly imaginative, collaborative, and exact.
Deep learning is used by ChatGPT, an intelligent programme, to replicate human reactions to conversational text inputs.
ChatGPT-4, the fourth generation of the chatbot, works specifically with massive amounts of digital data to develop the ability to provide precise results equivalent to those of humans.
The model is currently available to chatGPT Plus subscribers who pay $20 per month. You can also connect with chatGPT-4 if you have the most recent Bing.
Bing is now using the OpenAI GPT-4 model, which underpins chatGPT-4, according to Microsoft.
