April 2, it was announced that OpenAI would lower the threshold for using its AI chatbot ChatGPT, allowing users without an account to use it, albeit with some limitations.
Starting today, users in some regions will no longer need to log in to access chat.openai.com to directly interact with ChatGPT, and this free login access will be gradually opened in other regions around the world. Free login users can directly have a conversation with ChatGPT using the same model as logged-in users.
It is noted that free login users cannot enjoy some advanced features, such as saving or sharing chat logs, using custom commands, etc. These features usually require association with a permanent account.
More importantly, free login users can choose to opt out of the option of using chat logs for model training (this may be one of the main purposes of OpenAI to offer free login). Users only need to click on the question mark icon in the lower right corner, then enter “Settings” and turn off the relevant function.
However, the free login version of ChatGPT will have “more stringent content policies.” What does this specifically mean? In response, an OpenAI spokesperson gave a long but ambiguous explanation: “The free login experience will benefit from existing built-in security measures in the model, such as refusing to generate harmful content. In addition to these existing measures, we have also implemented additional safeguards for the types of inappropriate content that may occur in the free login experience. We have considered the potential ways in which the unlogged-in service may be abused, and combined with our understanding of the capabilities of GPT-3.5 and the risk assessment conducted.”
In other words, it is not yet clear exactly what these more stringent content policies are. As a large number of users flock to try out this new feature, we will soon find out the answer. “We recognize that further iterations may be needed and welcome user feedback,” the OpenAI spokesperson said.
In addition, for abuse, OpenAI also gave an ambiguous reply: “We also carefully considered how to detect and prevent abuse of the free login experience. The team responsible for detection, prevention, and response to abuse was involved in the design and implementation of the entire free login experience and will continue to support the design.”
The lack of specific information in OpenAI’s response may mean that they, like us, are unable to anticipate how users will use this free login feature, so they can only adopt a passive response strategy.
It is currently unclear which regions or user groups will be the first to get free login to ChatGPT, but the plan has been launched from today.
+ There are no comments
Add yours