We’re always committed to keeping your data as secure as possible. For us, it’s an inseparable part of our business model, and we prioritize it at every step.

This latest change is no different.

When creating short links, Short.io offers the option to use AI-generated slugs pulled directly from the content on your linked sites.

But up until now, this option was turned off by default.

Our previous approach

Why?

Well, because previously, data would pass through us, through Amazon Web Services, and finally onto OpenAI.

This meant that, while your data was secure for two-thirds of the process, it eventually found its way onto OpenAI’s servers.

Despite ChatGPT being a great model, it was still a third party in this transaction. And so we wanted to make sure that users can only opt in to having their data stored on OpenAI’s servers by choice.

Because we knew this would be an issue for some users, we turned off the option by default.

Improved data privacy

If you take a look the next time you create a link, you’ll see that AI slugs are now turned on by default.

So why the change?

In an effort to improve the privacy of our users, we switched from OpenAI’s ChatGPT to an in-cloud implementation of Anthropic’s Claude Haiku and Sonnet models.

For Mainland China, we adhere to the same standard, the only change being the use of Qwen instead of Claude.

These changes both come with one very important difference — improved data privacy.

While the models are still managed by AWS, no data actually reaches a third party this time, and this includes Anthropic and Alibaba themselves.

It all passes through a simpler pipeline — you, us, AWS. 

Even better — with this change, non-AI generated data has the same level of privacy as AI-generated data.

And if you still feel uncomfortable with this process, don’t worry — you can still turn it off.

Why do it?

We don’t compromise on privacy. That’s been an integral rule since our first lines of code.

While providing democratized access to short links is our main focus, we believe that’s only half a service.

Unless your links, personal information, and business remain entirely your own, we haven’t done our job.

In order to achieve this, we’re always on the lookout for ways to earn your peace of mind.

While there’s nothing wrong with OpenAI in principle, removing one extra step in our data storage pipeline and using in-cloud models ourselves means that we can simplify the flow of information.

It’s this streamlining that insulates your data from third parties and keeps it encrypted and safe throughout.

This is just one step we’ve taken toward the privacy of our users.

And of course, there’s many more to come.