Michael Minogue is running for governor of Massachusetts. Right now he's also running against a fake version of himself. A super PAC built a video wearing his face and voice, saying things he never said. And in Massachusetts, almost no law exists to stop it. Think about what that means for you. Not a politician, you. A convincing fake video of your kid or your parent can land online tonight. It spreads before anyone reports it. The damage happens in real time. The law, if one even applies, shows up much later.
The Boston Globe called it a deepfake free-for-all. Forty-five other states have passed at least one deepfake law. Massachusetts lawmakers went home without one.
The scariest number in this story is zero. That's how many reasons you have to assume the law catches up before the damage is done.
If the law won't act fast, the question becomes who's collecting your face and who's asking permission first.
You open your iPhone, tap People, and there's your daughter. Your dad from Christmas years ago. It's a feature you probably love. But to sort those photos, your phone measured the geometry of every face you've ever photographed. It built a faceprint for each person, a mathematical fingerprint of their face. In Illinois, doing that without written consent is illegal. And it may have happened six and a half million times.
As reported by PetaPixel, that exposes Apple to a potential thirty-two billion dollar lawsuit. Six and a half million residents, five thousand dollars each. A federal appeals court just cleared it for trial.
Apple didn't build this to harvest your face. They built it because people love it. But good intentions don't answer the real question. Did anyone ask you first?
Machines measure faces confidently. But confidence and accuracy aren't the same thing.
An investigator uploads a photo. The software returns a number. Zero point nine four. They read it as ninety-four percent certain, and start drafting an arrest warrant. That number sounds airtight. It isn't. Depending on settings the investigator can't even see, that same score could be wrong one time in a million, or one time in ten. And the software won't tell you which. If a facial match ever touches your job, your travel, or a legal case, this matters to you directly.
According to the Center for Democracy and Technology, systems marketed as ninety-nine percent accurate can drop thirty to forty points on real-world surveillance footage. And errors aren't shared equally.
False positives hit women and Black Americans hardest, up to a hundred times more often. A match is a lead, not a verdict. The algorithm hands you a number and walks away.
Three stories, one thread. A face is now data, whether it's faked, harvested, or scored by a machine. And in every case, the technology moves first. The law, the consent, the human check, they all arrive late. The gap between them is where real people get hurt.
Links to every story and today's podcast deep-dives are in the description. See you next time.
