Tech | Visa | Scholarship/School | Info Place

ChatGPT no longer requires an account – but there’s a catch

OpenAI is making its flagship conversational AI available to everyone, even those without an account. However, it won’t be the exact same experience – unless you opt out, of course, all of your chats will still make it into their training data.

Starting today, in some markets and gradually rolling out to other parts of the world, visiting will no longer require you to log in – but you can still log in if you wish. Instead, you will talk directly to ChatGPT, which will use the same model as the logged in user.

You can chat as much as you want, but be aware that you don’t get the exact same feature set as someone with an account. You won’t be able to save or share chats, use custom directives, or other content that would normally be associated with a persistent account.

That said, you can still choose not to have your chats used for training (which, one suspects, undermines the whole reason companies do this in the first place). Just click the little question mark in the lower right corner, then click Settings, then disable the feature. OpenAI includes this helpful gif:

Image Source: open artificial intelligence

What’s more, this additional free version of ChatGPT will have “slightly stricter content policies.” what does that mean? I asked a question and got a lengthy but largely meaningless response from a spokesperson:

The logout experience will benefit from existing security mitigations already built into the model, such as denying the generation of harmful content. In addition to these existing mitigations, we have implemented additional protections specifically designed to address other forms of content that may not be suitable for the logout experience.

Based on our understanding of the functionality of GPT-3.5 and the risk assessment we have completed, we have considered potential ways in which logged-out services could be used in inappropriate ways.

So…really, no idea what these more restrictive policies are exactly. No doubt we’ll find out about this new offering soon, as Landos floods the site. “We recognize that additional iterations may be needed and welcome feedback,” the spokesperson said. They will have it – and in plenty!

On that note, I also asked if they had any plans to deal with the unprecedented scale of attempts to abuse and weaponize the model that will almost certainly occur. Just imagine: a platform whose use causes billionaires to lose money. After all, inference is still expensive, with even the elegant low-lift GPT-3.5 model requiring power and server space. People will go to great lengths to hone it.

In response to this threat, they also gave a lengthy non-answer:

We’ve also carefully considered how to detect and prevent abuse of the signed-out experience, and the teams responsible for detecting, preventing, and responding to abuse were involved in the entire design and implementation of this experience and will continue to inform its design moving forward.

Note the lack of anything resembling specific information. They probably don’t know as much as anyone else what people are letting this happen, and have to be reactive rather than proactive.

It’s unclear which regions or groups will get the super-free ChatGPT first, but it starts today, so check back regularly to see if you’re one of the lucky ones.

#ChatGPT #longer #requires #account #catch

Leave a Reply

Your email address will not be published. Required fields are marked *