top of page
  • Writer's pictureBig Data Ben

Gandolf, the friendly AI security bot

October 12, 2023

Have you ever wondered how some apps can write amazing texts for you, just by giving them a few words or a topic? Well, these apps use a special kind of artificial intelligence (AI) called large language models (LLMs), which are like super-smart machines that can understand and create human language.

Sounds cool, right? But what if someone tries to trick these machines into doing something bad, like giving away secret information or breaking into your accounts? That's where Lakera comes in. Lakera is a new company that helps protect LLMs from malicious prompts, which are like sneaky commands that hackers use to fool the machines.

Lakera has developed a huge database of millions of examples of how hackers can try to attack LLMs, and they use this data to train their own AI system to detect and prevent these attacks. Lakera also offers an easy way for developers to integrate their security solution into their apps, so they can use LLMs safely and confidently.

Lakera was founded by a team of experts who have worked on AI for aerospace and healthcare, where security is very important. They also created a fun game called Gandalf, where you can try to hack an LLM yourself and see how it defends against your prompts. You can play Gandalf here.

Lakera is one of the first companies to focus on AI security, and they have already attracted some big customers and investors. They believe that AI has a lot of potential to improve our lives, but only if we use it responsibly and securely. That's why they are on a mission to make AI safe for everyone.

If you want to learn more about Lakera and their products, you can visit their website or read the original article that inspired this blog post. I hope you enjoyed this blog post and learned something new about AI. Thank you for reading!

9 views0 comments
bottom of page