Americans Lost $893M to AI Scams in 2024, FBI Says

The FBI's 2025 Internet Crime Report has finally put a price tag on something most people in security have been muttering about for a year: AI scams work, and they work at scale. Americans filed 22,364 complaints involving AI-driven fraud in 2024, with reported losses of $893,346,472. And that's the floor, not the ceiling, because a huge share of victims never report.

What's actually new here

The categories of crime are old. Romance scams, fake kidnapping calls, dodgy investment pitches, impersonation of tax officials and police, business email compromise. None of these were invented by generative AI. What AI has done is remove the tells.

The bad grammar is gone. The strange accent is gone. The obviously photoshopped image is gone. In its place: voice cloning that needs only a few seconds of source audio, deepfake video good enough to survive a quick call, and AI-generated scripts that match the tone of a real executive or a real government agency.

Michael Machtinger, deputy assistant director of the FBI Cyber Division, told the Wall Street Journal that AI-generated fraudulent communications now look convincing even to people who know what to look for. That is the bit worth sitting with. The advice we've been giving for twenty years, look for spelling mistakes, listen for the off-sounding accent, check if the photo looks weird, is quietly being retired by the criminals themselves.

Government impersonation, upgraded

The clunky IRS gift card call is being replaced by something more polished. Spoofed caller ID showing a real agency number. Lifted branding on follow-up emails. Cloned audio, and in some cases cloned video, of named public officials. For an older victim who has been told for years to trust callers who can cite the right department, this is a nasty shift.

Business email compromise gets a voice

BEC was already the most expensive category of cybercrime in dollar terms before AI joined in. Now the report flags cases where a finance team received not just an email from the CFO but a follow-up voice note, or a quick video message, in the CFO's actual voice and manner. Tens of millions have already moved on the strength of those calls.

What changes in practice

The uncomfortable bit is that the practical advice has narrowed to one line:

• A familiar voice is not proof of identity.
• A familiar face on a video call is not proof of identity.
• Anything urgent and financial gets verified through a separate, known channel before anyone moves money or shares credentials.

That used to be a belt-and-braces extra step for high-value transactions. It is now the baseline for any request that pressures someone to act fast. The scammers have caught up with the surface signals. The only thing left that they cannot fake cheaply is a second channel the victim controls.