OpenAI, ChatGPT is extending access to its leading conversational AI to everyone, even those who haven’t registered an account. However, the experience won’t be identical. Additionally, all conversations will continue to contribute to their training data unless users choose to opt out.
Beginning today, in select markets and expanding globally over time, users visiting chat.openai.com will no longer be prompted to log in. Instead, they’ll be immediately immersed in conversation with ChatGPT, utilizing the same model as registered users.
Feel free to chat as much as you like, but keep in mind that you won’t have access to all the features available to users with accounts. Saving or sharing chats, utilizing custom instructions, and other functionalities typically tied to a persistent account won’t be available.
However, you still retain the choice to opt out of your conversations being utilized for training purposes, although this option may contradict the primary motivation behind OpenAI’s decision to implement this change. You can simply click on the small question mark icon in the lower right-hand corner, proceed to “settings,” and disable the feature there. OpenAI provides a helpful GIF demonstrating this process.
More significantly, this no-cost version of ChatGPT will implement “slightly more restrictive content policies.” But what exactly does that entail? Upon inquiry, a spokesperson provided a verbose yet largely ambiguous response.
The unsigned experience will enjoy the same safety measures embedded within the model, including the refusal to generate harmful content. Moreover, we’re introducing extra safeguards tailored to address other types of content that might not be suitable for an unsigned experience.
Our considerations encompassed various scenarios where an unsigned service could potentially be misused. These considerations were informed by our comprehension of GPT-3.5’s capabilities and assessments of potential risks that we’ve conducted.
So… there appears to be little clarity on the specifics of these stricter policies. However, it’s likely that a better understanding will emerge soon as a wave of curious individuals descends on the site to test out this new offering. The spokesperson acknowledged, “We recognize that further refinement may be necessary and welcome feedback.” And feedback they shall receive — in abundance!
On that note, it was also inquired whether there are any strategies in place to manage what will likely be attempts to exploit and misuse the model on an unprecedented scale. Just imagine: a platform whose use could potentially result in financial losses for a billionaire. After all, running inferences still comes at a cost, and even the optimized, resource-efficient GPT-3.5 model requires power and server space. People will undoubtedly push it to its limits.
For this threat they also had a wordy non-answer:
Furthermore, we have meticulously pondered methods to detect and prevent misuse of the unsigned experience. The teams tasked with identifying, thwarting, and addressing abuse have been integral throughout the design and execution of this experience. Their involvement will persist in shaping its development moving forward.
Notice the lack of anything resembling concrete information. They probably have as little idea what people are going to subject this thing to as anyone else, and will have to be reactive rather than proactive.
It’s not clear what areas or groups will get access to ultra-free ChatGPT first, but it’s starting today, so check back regularly to find out if you’re among the lucky ones.
