Since the release in May 2020, the too high-performance text generation model "GPT-3" has been used only for limited developers due to abuse risks.The developer Openai has recently unlocked the API waisting list and opened the GPT-3.
Currently, everyone can use content generation and automatic generation of code by CODEX.However, it is necessary to follow the rules to prevent misuse.
Openai has taken various safety measures in the public's release.Content guidelines that clearly specify the available content categories, and content filters to comply with this.The developed app requires a review by Openai before the release, and the API key will be canceled without this process.
By the way, according to the content guidelines, political content is not available at the moment, besides hate, adult, spam and malware.In addition, the prospect of approval is shown for each use case, and it seems that it is generally approved in the applications that summarize short sentences collected by customer support.On the other hand, in the application for the end user to summarize the desired sentences, the approval is a case -by -case because there is a possibility of generating non -safe content.
Openai has shown intention to prevent GPT-3 abuse through monitoring and development support in addition to providing and reviewing safety tools such as content filters.
With the improvement of the system, such as the integration of the content filter, the use will be more flexible, and it is likely that the opportunity to see the GPT-3 sentences will increase.
Source: Openai's API now Available with no Waitlist/ Open AI
(Sentence / Yo Yamada Road)