Deepfakes are increasingly used in fraud, most notably in corporate wire-fraud schemes where attackers impersonate executives on live video or voice calls to authorize transfers — one documented case cost an engineering firm .6 million in a single day. The countermeasure that has repeatedly beaten even sophisticated deepfakes is out-of-band verification: confirming a request through a separate channel, such as calling back on a known phone number, rather than trusting how convincing a call looks or sounds. Regulation like the EU AI Act increasingly requires disclosure and machine-readable watermarking (often via C2PA) for AI-generated synthetic media.