In this episode of What is, let's understand what is Jailbreaking an AI?
Jailbreaking an AI is a process by which one can bypass the guardrails of a generative AI models, enabling it to respond to queries which it otherwise would not. Malicious actors, are known to jailbreak generative AI using prompt injection, evasion, and model manipulation. Do you remember the ChatGPT alter ego, Dan, which became popular in its early days? Well, it has been patched since then, but that is a classic example of an AI jailbreak.
What is? is an educational video series hosted by Digit Expert Satvik Pandey ( @satvikp ) In What is? Satvik takes the most complex tech jargon and explains it to you in 60 seconds! Catch What is? on Digit’s social media platforms on Sunday at 9AM every week.
17