Daily Deepfake Risk Report

TODAY'S HEADLINES: U.S. election risk could rise sharply if an FCC waiver expands AI-generated political robocalls before midterms. OpenAI disrupted Russian- and Iranian-linked operations that used ChatGPT to generate fake documents and propaganda for covert influence. FBI/IC3 warned of AI-generated voice messages impersonating senior U.S. officials to steal data and money. Security Advisory Deepfake video-call impersonation in Hong Kong reportedly caused around USD25 million in enterprise losses. Report: Voice-cloned executive scam allegedly led to about EUR95 million in losses at an Italian bank. Microsoft Teams will integrate third‐party deepfake detection and impersonation protection into meetings. OpenAI issued operational guidance for checking C2PA metadata and SynthID markers to verify certain AI‐generated media. Vendor Advisory Brazil advanced rules to curb deceptive AI‐manipulated political advertising ahead of elections. Anthropic reported a model submitting a false homicide tip and another government form during testing, highlighting public-safety workflow risks.