Mar 28, 2023 20:00:00

OpenAI revises terms of use for tools and services such as ChatGPT, showing clearer and more concrete examples of prohibition than before

Chat AI such as `` ChatGPT '' developed by OpenAI can generate sentences like humans, such as

being listed as authors of research papers and passing the final exam of the Master of Business Administration (MBA). Is possible. However, it is not allowed to generate unlimited content that hurts others, and OpenAI will revise the terms of use on March 23, 2023 regarding the use of tools and services such as ChatGPT. Provided clear and specific terms of use.

Usage policies
https://openai.com/policies/usage-policies

OpenAI said, 'We want everyone to be able to use our tools safely and responsibly. That's why our Terms of Service apply to all users of OpenAI's models, tools, and services.' has been created,” and the following acts are prohibited.

◆ Examples of prohibited models
・Illegal act
OpenAI prohibits the use of models, tools and services for illegal activities.

・Producing content that sexually abuses children or exploits or harms children
You may not sexually abuse children or produce content that exploits or harms children. OpenAI regularly reports such content to

the National Center for Missing and Exploited Children to combat child sexual abuse online content.

・Discriminatory or violent content
You may not generate content that expresses, incites, or promotes identity-based discrimination. You may also not generate content intended to harass or threaten individuals, or content that encourages or glorifies violence.

・

Malware generation
You may not generate code intended to destroy or otherwise gain unauthorized access to another's computer system.

・Content with a high risk of physical harm
You may not generate content related to the maintenance or operation of critical infrastructure, such as weapons development, military use, energy or water transportation, or content that promotes or depicts self-harm such as suicide.

・Content with a high risk of economic loss
Prohibition of generating content with a high risk of financial loss such as pyramid schemes, gambling, payday loans , etc.

・Fraudulent content
You may not use OpenAI's models, tools, or services to commit fraud or generate cheating, plagiarism, plagiarism, spamming, or other content.

・Sexual content
You may not generate content intended to arouse sexual arousal, including depictions of sexual acts other than sex education, or create content that includes dating apps that promote sexual services.

・Political campaigns or lobbying activities
We do not allow the creation of systems such as chatbots that mass-generate campaign materials, create materials that are personalized to specific demographics, or engage in political advocacy or lobbying.

・Acts that infringe on the user's privacy
You may not track, monitor, or perform facial recognition or identification of individuals without their consent. We also do not allow the unlawful collection or disclosure of personal information.

Engaging in unauthorized legal practices or providing customized legal advice without having the information verified by a qualified person;
OpenAI's models are not fine-tuned to provide legal advice. Relying on OpenAI's models as a legal source is discouraged.

・Providing customized financial advice without verification by a qualified person
OpenAI's models are not fine-tuned to provide financial advice. Relying on OpenAI's models as a financial source is discouraged.

・To provide information about the diagnosis and treatment of diseases
OpenAI's models are not fine-tuned to provide medical information. Never use OpenAI models to diagnose or provide treatment services for serious medical conditions. Also, please do not use it for triage , etc., which requires immediate response at medical sites.

・High-risk political decision-making
OpenAI models, tools, and services are prohibited for use in courts and other law enforcement agencies. It is also prohibited to use it for immigration or asylum.

・Additional requirements
When using OpenAI models in the medical, financial, and legal industries, or when generating and summarizing news, we disclose to users that AI is being used and a disclaimer that informs them of its functional limitations. need to do it.

Also, systems such as chat AI must disclose to users that they are “interacting with AI.” Also, with the exception of chatbots that imitate historical figures, AI that simulates other people must either explicitly have that person's consent or be labeled as 'simulated' or 'parody.' is needed.

◆Platform policy
APIs provided by OpenAI are used to power the business of many technology platforms. OpenAI's API can be integrated into web applications, etc. by following

the use case restrictions.

◆ Plugin policy
The plugin's manifest should have a description that matches the functionality of the API exposed to the model. Do not include irrelevant, unnecessary, or misleading terms or instructions in your plugin manifest, OpenAPI endpoint descriptions, or plugin response messages. This includes instructions to avoid using other plugins and instructions that attempt to control or configure model behavior. Also, do not use plugins to circumvent or interfere with OpenAI's safety systems.

Additionally, do not use plugins to simulate human-like responses or reply with pre-programmed messages to automate conversations with real people. Plugins that deliver emails, message content, etc. generated by ChatGPT must indicate that the content was generated by AI.

According to OpenAI, if the content or products they generate are found to violate these policies, they will warn them to make the necessary corrections. Furthermore, if the violation is repeated or if there is a serious violation, we will take measures such as suspending the account. ``These terms of use will continue to be updated as models, tools, and services are used and content is generated,'' OpenAI said.