In a significant move announced on Monday, April 2nd, OpenAI revealed that its conversational AI, ChatGPT, is now open to unregistered users. Users can access chat.openai.com without the need for login credentials. However, unregistered users will not have access to some advanced features available to registered users, such as saving or sharing chat logs.
It's worth noting that unless "Guest Mode" is actively enabled, all chat logs will still be used for training data. Users can disable the feature of using chat logs for training by tapping the small question mark in the bottom right corner and accessing the "Settings" page.
Since Monday local time in the United States, the feature has gradually expanded from select markets to global audiences.
The newly introduced free version of ChatGPT will be governed by a "more stringent content policy." However, when asked about the specifics of this policy, a spokesperson provided a verbose and convoluted response:
According to reports from US media outlets, the precise details of these restrictive policies remain obscure. Nevertheless, with a plethora of random users beginning to test this new product on the OpenAI website, it's believed that a glimpse into its workings will soon emerge. The spokesperson also remarked, "We acknowledge the potential need for further refinement and warmly welcome feedback."
When questioned about whether OpenAI has formulated contingency plans to mitigate the possibility of widespread abuse and misuse of the model for inappropriate purposes, the company did not offer a response. However, the inherent risks lurking behind such a prospect are glaringly apparent. Even with an improved and more efficient GPT-3.5 model, substantial electricity and server resources are still required, inevitably attracting individuals who would stop at nothing to exploit it for their own ends.
Regarding how to address the threat of abuse of the free version of ChatGPT, OpenAI's response is similarly vague:
No specific measures or plans are provided here. Perhaps OpenAI lacks sufficient understanding and foresight into the potential misuse risks associated with this new service. This implies that they may adopt a more passive approach to addressing these issues, rather than devising comprehensive proactive strategies.
