• Question: How do you prevent AI from being hacked and what is the how often does this happen?

    Asked by pawn535mars97 on 26 Feb 2026.
    • Photo: Arshia Keshvari

      Arshia Keshvari answered on 26 Feb 2026:

      If it has a backdoor anything can be hacked, that is if you tell them there is a backdoor or it’s immediately obvious regardless if it’s AI powered.

      Some of the most secure systems are ones that are proprietary or simply too sophisticated to use conventional tools to break into them.

Comments

  • Photo: aureliabrzezowska

    aureliabrzezowska commented on :

    It happens quite often, you can only prevent something once you know where the backdoor is and often is a panic once it is found to fix it. Common attacks against AI include prompt injections, jailbreaking (content safety focus), data poisoning, data extraction and many more!