NEW350+ rooms now have a live terminal, so you can learn by typing the commands, right in your browser.Learn in a live terminal. 350+ rooms.Try a lab♥ Codelivly is 100% free, no paywall. If it helped you, chip in to keep it running.100% free. Help keep it so.Support us
[>_] ROOM · Cybersecurity
Backdoored Models
Understand how attackers distribute and deploy backdoored AI models via supply chain attacks, and learn to detect and mitigate these hidden threats.
Room tasks
1What is a Backdoored Model & How Triggers Work
2How Attackers Embed Triggers During Training
3Techniques for Detecting Backdoor Behavior
Frequently Asked Questions
What is a backdoored AI model?
It is a model that performs normally until a specific trigger appears, then produces attacker-chosen output. The backdoor is hidden in the weights, so it is hard to spot through ordinary testing.
How do backdoored models spread?
They are often published to public model hubs or shipped as fine-tuning adapters such as LoRA. Teams who download and deploy them without verification inherit the hidden behavior.
Can backdoors be detected?
Detection is difficult but possible using trigger scanning, activation analysis, behavioral testing and provenance checks. This intermediate Codelivly room introduces those methods and is free to take.