Model Extraction
Model extraction is an attack in which someone recreates or steals a functional copy of an AI model by querying it repeatedly and training on its answers.
What is Model Extraction?
Extraction threatens intellectual property and can help attackers find weaknesses. Rate limits and monitoring of unusual query patterns are common protections.
How it works
The attacker sends many inputs and uses the outputs as training data for a substitute model.
← All Artificial Intelligence & Superintelligence definitions · ← Back to the A–Z definitions