OpenAI has made its flagship conversational AI accessible to all, including those who haven’t created an account. However, there’s a catch—your experience won’t be the same, and all your conversations will still contribute to their training data unless you choose to opt out.
Starting today in select markets and gradually expanding worldwide, visiting chat.openai.com will no longer prompt you to log in—although you still have the option to do so. Instead, you’ll immediately engage in conversation with ChatGPT, utilizing the same model as logged-in users.
While you can chat freely, it’s important to note that you won’t have access to the full range of features available to users with accounts. Functions like saving or sharing chats, using custom instructions, and other account-associated features won’t be available.
However, you still retain the choice to opt out of having your chats used for training purposes—an option that may seem counterintuitive to the company’s motives for this change. To do so, simply click the small question mark icon in the lower right corner, navigate to “settings,” and disable the feature. OpenAI provides a helpful GIF demonstrating this process.
Of greater significance, this additional free version of ChatGPT will adhere to “slightly more restrictive content policies.” What exactly does that entail? When queried, a spokesperson provided a verbose yet largely vague response:
“The experience for users who haven’t logged in will be enhanced by the model’s existing safety measures, including its ability to refrain from generating harmful content. Additionally, we are implementing supplementary safeguards tailored to address other types of content that may not be suitable for users without accounts.
“We have additionally meticulously examined strategies to recognize and hinder abuse of the platform for individuals who haven’t logged in, leveraging our comprehension of GPT-3.5’s abilities and performing risk evaluations.”
In essence, the specifics of these stricter policies remain unclear. Undoubtedly, as a wave of new users flock to the site to explore this new offering, clarity will emerge.”The spokesperson added that we recognize the potential need for further refinement and encourage feedback.” Feedback will indeed be plentiful.
On a related note, I inquired about OpenAI’s strategy for dealing with the inevitable attempts to exploit and manipulate the model on an unprecedented scale. Consider the potential ramifications: a platform whose usage results in financial losses for individuals. Given the computational resources required for inference, even the streamlined GPT-3.5 model demands power and server space. Users will inevitably push it to its limits.
Yet again, the response was evasive:
“We have also extensively reviewed methods to detect and prevent misuse of the service for users who haven’t logged in, based on our understanding of GPT-3.5’s capabilities and risk assessments.”
Notably absent is any concrete information. It appears that OpenAI may have limited insight into the potential challenges this service will face and will likely adopt a reactive rather than proactive stance.
The specific demographics or groups granted access to this enhanced free version of ChatGPT first remain unclear. Nevertheless, the rollout begins today, so regular visits may reveal whether you are among the fortunate initial users.
