Let’s talk about why suspicion does not scale.
Picture a magistrate in a dusty South African courtroom. Two neighbors disputing a verbal agreement. One pulls out a voice recording that sounds exactly like the other man agreeing to the terms. The defendant swears it’s fabricated. The magistrate stares at the file, shrugs, and throws the case out. Both men leave angrier than when they arrived. Multiply this scene by every family chat, every supplier contract, every urgent executive call. That’s where we’re headed.
The machines have made forgery cheap. What used to require expensive studio equipment now takes a decent prompt and five minutes. Voices clone in seconds. Videos stitch themselves together with physics that look better than real life. And our first instinct, the reasonable one, is to trust less. Double-check everything. Assume it’s synthetic until proven otherwise.
This feels like vigilance. It isn’t.
Suspicion at scale breaks the operating system of human life. You cannot run a business if every email might be forged. You cannot maintain a friendship if every voice note requires forensic analysis. You cannot raise children in an environment where they learn that nothing said by an adult voice can be taken at face value. The exhaustion compounds. People withdraw. Institutions lose whatever thin legitimacy they had left. And the real liars? They thrive in the noise.
This is what some call the liar’s dividend. The more doubt we spread, the easier it becomes for genuine deception to hide. The fog benefits the worst actors. We’ve seen early versions of this in politics and news. Now it’s coming for everyday trust.
The deeper problem is what it does to us. Humans are not built for permanent skepticism. Our judgment, our taste, our sense of enough — these things grow in soil that includes reasonable trust. Strip that away and you don’t get sharper humans. You get paranoid ones. Brittle ones. People who stop taking risks, stop reaching out, stop believing that a stranger’s word or a friend’s commitment can mean anything.
I keep returning to the court example because it’s so ordinary. Not some high-profile deepfake scandal in New York or London. Just two people who used to solve problems with a handshake and a shared understanding that words have weight. That weight is disappearing. Not because humans changed. Because the infrastructure around our words stopped anchoring them to anything real.
The fix people reach for first is more detection. Better AI to catch the fake AI. Arms race logic. It won’t work. The generation tools improve faster than the detectors, and the average person cannot be expected to run confidence scores on every interaction. That path turns all of us into unpaid security analysts. It’s not sovereignty. It’s servitude with extra steps.
What must survive is a different principle. Proof should travel with the thing itself.
Not centralized gatekeepers we beg for validation. Not post-hoc analysis that clever fakes will always outrun. But the ability for a human to sign their creation at the moment it’s made — a cryptographic tether, a verifiable credential, a provenance layer that says this came from this person, at this time, under these conditions, without handing over their entire life history.
This is where consent stops being a clickbox and becomes infrastructure. Where identity becomes selective proof instead of total profile. Where we stop asking the machine to simulate trust and start demanding it carry verifiable human intent.
It sounds technical because it is. But the stakes are not. They’re the difference between a world where a parent can receive a genuine message from their child and know it, versus living in permanent low-level dread that nothing is real. They’re the difference between commerce that can move at human speed and commerce that requires layers of lawyers and third-party escrow for every minor transaction. They’re the difference between relationships that can survive distance and relationships that slowly dissolve under the acid of reasonable doubt.
We’ve watched this pattern before. The internet made information abundant and created the need for better discovery. Social platforms made connection frictionless and created the need for better identity and moderation. Generative systems now make impersonation trivial. The pattern is clear. Each leap in machine capability forces us to make human agency more explicit, not less.
The builders who understand this will win the next decade. Not the ones shipping faster slop. The ones who design systems where the human stays in control of their signal. Where provenance is default, not an afterthought. Where consent has teeth because it’s baked into the protocol instead of hidden in terms of service nobody reads.
This isn’t anti-technology. It’s pro-human infrastructure.
The machines will keep getting better at pretending. That’s what they do. Our job is to make sure reality stays expensive to fake. Not by yelling at the clouds or adding more detectors. By building the quiet plumbing that lets truth travel attached to its source.
Because at some point the question stops being “can we detect the fake?”
It becomes “can we still look each other in the eye — across screens, across distances, across time — and know that what we’re seeing carries the weight of a real person with real stakes?”
That weight is worth protecting.
Suspicion doesn’t scale.
Verifiable human provenance does.
And the choice between those two futures is being made right now, in the architecture we decide to ship.