Last updated:
An AI model that is trained on a vast amount of data and has high-level performance that could be used for both civilian and harmful or military purposes.
While not a formal tier in the EU AI Act's risk pyramid, dual-use concepts often overlap with systemic risk GPAI models. These models possess capabilities (like advanced chemical synthesis or cyberattack planning) that require red-team testing and safety reporting to the EU AI Office to prevent their misuse in national security contexts.
Real world example:
A tech giant releases a model that can write both poetry and functional malware; because it has dual-use potential the provider must share red-team safety results with government regulators.
