Summary

  • DeepSeek AI’s new R1 model is taking the world by storm, but it may not be all that secure.
  • Wiz Research found that the product left data exposed, revealing tons of sensitive information.
  • The issue has now been fixed, but it’s unclear just how many people had access.

If you’re someone that’s been in the tech space for the past few years, it’s been a pretty wild ride since the introduction of artifical intelligence to the mainstream. While OpenAI really got the the ball rolling with the public release of ChatGPT, plenty of other companies soon followed, introducing its own tools and services in order to compete.


Related


Microsoft wants to know if DeepSeek obtained data from ChatGPT

OpenAI thinks DeepSeek ripped them off



3

For the most part, the competition has been playing catch up, with brands like Google and Microsoft playing second fiddle to OpenAI. But that doesn’t mean that these same companies aren’t making strides when it comes to developments in AI. Of course, these aren’t the only players in the space, with brands from around the world still trying to make their mark.

The hole has now been plugged

Over the past week, China’s DeepSeek AI has been seeing some heavy traction, with the company showing off its new R1 model that is completely free to use, while also being at the cutting edge. The big story has been that this new model isn’t all that expensive to train or maintain, but some have raised questions about how DeepSeek is accomplishing these feats.

While it’s still unclear at this time, companies are racing to see what makes R1 tick, and because of that, researchers have discovered that the DeepSeek may not be all that secure. The Wired first reported about the security hole with the help of Wiz Research, who originally discovered the issue that leaves data exposed (via The Verge).

A little digging by researchers found that it wasn’t all that hard to access the data and even more alarming is that authentication was not necessary. Wiz Research shares that “this database contained a significant volume of chat history, backend data and sensitive information, including log streams, API Secrets, and operational details.”

And while it’s unclear whether this loophole was exploited by others, the good news is that the issue has now been fixed. Of course, for these reasons and more, it’s always good to be cautious about what you type into these types of products, because you don’t know where the data will go and how it will be used. Some apps and services even warn users not to enter any sensitive data into prompts.

With that said, this probably isn’t the last time that we’re going to hear about DeepSeek. Its introduction could be the start of something new, and we could even see some big changes in the space, especially when it comes to competitors like OpenAI, Nvidia, Microsoft, and others. It’s definetely worth keeping an eye out for because things could really start to get interesting.