Fake Biden robocall tells voters to stay home

Imagine picking up your phone before a crucial primary, and the sitting president tells you to skip Tuesday's vote.

That scenario just played out for real in New Hampshire. A robocall featuring a cloned voice of President Joe Biden made the rounds, urging Democrats not to bother writing in his name in the state's presidential primary. The state attorney general's office immediately stepped in, calling it an unlawful attempt to suppress votes, and disinformation researchers confirm the recording is almost certainly an AI deepfake.

Clone a politician in minutes

The audio, first obtained by NBC News, sounds remarkably like Biden, though experts pointed out some telltale glitches. Lindsay Gorman at the German Marshall Fund noted that the cadence turns clipped and robotic toward the end. Ben Colman, CEO of detection firm Reality Defender, said while nobody has ground truth, all signs point to manipulation.

The real problem isn't that the fake was flawless. It is that it was effortless.

Voice-cloning apps and websites are widely available for a few bucks or completely free. Because public figures leave behind thousands of hours of clean audio samples, anyone with an internet connection can build a passable clone of a world leader in minutes. Senator Richard Blumenthal warned that a total neophyte can master this technology instantly.

Old playbook, dangerous new weapon

Voter suppression via robocall is not new. Political operatives have long dialed voters with misleading polling locations or fake deadlines.

As Mekela Panditharatne of the Brennan Center for Justice pointed out, voice-generation AI simply turbocharges that legacy playbook. The scam becomes infinitely cheaper, more scalable, and significantly harder for an average person to dismiss on an unexpected phone call.

Meanwhile, Washington is caught flat-footed. Federal law prohibits deceptive efforts to block people from voting, but rules around deceptive AI remain virtually nonexistent. The Federal Election Commission has faced mounting pressure to regulate deepfakes in political ads, yet a spokesperson admitted the agency hasn't even determined whether it will begin formal rulemaking.

Why it matters

This incident crosses a critical line: weaponized AI in elections has officially moved from a theoretical warning to a live operational attack on voters.

For regular citizens, defense is nearly impossible. Spotting unnatural eye movements in a deepfake video is hard enough, but catching an unnatural cadence in a crackly, ten-second phone call while rushing to work is unrealistic. The threshold to deceive someone into staying home is far lower than convincing them to switch parties.

In our view, the response from regulators looks embarrassingly slow. While bad actors are already deploying automated clone calls at scale, agencies like the FEC are still debating whether to start the process of writing rules. Unless enforcement catches up to consumer-grade AI tools immediately, deepfake suppression calls will become standard campaign warfare everywhere.