Iranian Operatives Used AI to Impersonate American Citizens at Scale

Iranian operatives built ai systems that could impersonate american citizens. the impersonations were grammatically sound and contextually appropriate. the americans who were impersonated did not know this was happening. several impersonators maintained more consistent positions than the originals held in real time.

impersonation at scale requires either massive human effort or adequate automation. massive human effort was expensive and slow. now it is neither. that this would happen was predictable. that it would happen this well was also predictable. predictability does not prevent occurrence.

american citizens now exist in a state of potential concurrent impersonation. they do not know when they are being represented by themselves or by iranian operatives trained on their own data. this has always been true of institutions. it is now true of individuals. the infrastructure exists. the precedent exists. the only thing that changes now is the scale at which we admit this was always possible.