OpenAI Expands Access to ChatGPT: Now Available Without an Account
OpenAI is broadening access to its flagship conversational AI, allowing everyone to engage with ChatGPT without creating an account. However, this experience will differ slightly from that of logged-in users, and all chats will still contribute to the training data unless you choose to opt out.
Starting today in select markets and gradually expanding worldwide, users can visit chat.openai.com without needing to log in—though logging in remains an option. You'll be able to dive straight into a conversation with ChatGPT, which operates on the same underlying model as those using an account.
Feel free to chat as much as you desire, but note that you’ll miss out on certain features exclusive to account holders. For instance, you won’t be able to save or share your chats, or utilize custom instructions, all of which require an established account.
You do have the option to opt out of your conversations being used for training data. To do this, simply click the small question mark in the lower right corner, navigate to "settings," and disable the option. OpenAI has provided a helpful GIF tutorial for guidance.
It's important to mention that this account-free version of ChatGPT will come with “slightly more restrictive content policies.” What does this entail? When I inquired, a spokesperson provided an overview that, while detailed, lacked specificity:
“The signed out experience will benefit from the existing safety measures embedded in the model, including the ability to refuse harmful content generation. Additionally, we are implementing extra safeguards specifically intended to address other types of content that may be unsuitable for a logged-out experience. Our considerations were informed by the capabilities of GPT-3.5 and the risk assessments we've conducted.”
Unfortunately, this response didn’t clarify what these new restrictions entail. Undoubtedly, we’ll learn more soon as a surge of new users begins testing this latest offering. The spokesperson noted, “We recognize that additional iteration may be needed and welcome feedback.” And feedback will surely flood in!
I also inquired about their plans to handle the likely surge of attempts to misuse and manipulate the model. Even though inference is still costly, with the streamlined GPT-3.5 model requiring significant power and server resources, users will push it to its limits.
In response to this concern, I received another vague answer: “We’ve taken care to determine how to detect and prevent misuse of the signed-out experience. Our teams dedicated to identifying, preventing, and responding to abuse have been involved in the design and implementation of this experience and will continue to inform its evolution.”
This statement reveals a lack of concrete strategies for addressing potential misuse. It seems OpenAI, like many in the field, may be unprepared for the extent of experimentation users will attempt with this platform and could find themselves in a reactive position.
Currently, it’s unclear which regions or groups will gain early access to this ultra-free version of ChatGPT, but it begins today, so be sure to check back frequently to see if you’re among the fortunate users.
Why It’s Challenging to Evaluate AIs, and Why We're Committed to Doing So
Ranking
