← FUURAA AI Frontier Library
EmergingPolicy, Law & GovernanceNow
Generative AI expands the attack surface
Generative systems add misuse, prompt-based evasion, privacy leakage and content-mediated risks to familiar machine-learning threats.
National Institute of Standards and Technology24 March 2025Reviewed 26 July 2026
FUURAA original conceptual visualWhat the evidence indicates
A concise reading of the source
Generative systems add misuse, prompt-based evasion, privacy leakage and content-mediated risks to familiar machine-learning threats.
FUURAA interpretation
Why this could matter
Threat models must cover user interaction and tool execution, not only the model file.
How to read this signal
A direction still taking shape
Multiple developments point in this direction, but timing, adoption and outcomes remain open.
