The Death of Truth Has Begun

There’s something bleakly funny about the age we’re walking into. For the first time in history, the future isn’t approaching us. It’s overtaking us. Most people are still trying to figure out how to change the ringtone on their phone while AI has already mastered impersonating their voice, their face, and their emotional weak spots with industrial precision.

We pretend we’re keeping up, learning the tools, adapting, “staying informed”, but the truth is that the pace has already broken free of human comprehension. Criminals saw this first. They industrialised deception while the rest of us were still trying to reset our online banking password. Deepfake fraud surged by 3,000% in a single year, according to Keepnet Labs, as criminals realised it costs roughly the price of a supermarket meal deal to clone someone’s voice or generate a fake executive on a video call. Global fraud losses are expected to hit $40 billion annually by 2027, according to Cobalt.io, and the line isn’t flattening, it’s curving upwards like it’s trying to escape the chart.

The real accelerant is the shift from human-speed crime to machine-speed, autonomous crime. We’ve entered the era of agentic AI, described by McKinsey as systems capable of reasoning, planning, acting, and adapting without supervision. For criminals, this is a dream scenario. An AI agent can identify targets, scrape their digital footprints, craft personalised phishing, adjust its tactics mid-conversation, and run thousands of campaigns in parallel, no fatigue, no oversight, no hesitation.

Even traditional phishing has undergone a grotesque evolution. The old “Dear Sir, I am Prince of Nigeria” era has been replaced with hyper-personalised AI emails boasting a 54% click-through rate, according to ACM. They are flawlessly written, psychologically tuned, and tailored to your job role, recent activity, or even your writing style.

Humans were never built to detect any of this, especially not older adults, who now represent the richest and softest targets. Fraud losses among seniors exceeded $3–5 billion, according to the FBI’s IC3 reports and supporting analysis from the National Council on Aging. But it’s not the financial loss that defines this threat. It’s the psychological cruelty. A cloned grandchild’s voice screaming for help creates a crisis that bypasses logic entirely. Victims describe the experience as “a nightmare,” according to the American Bar Association.

AI isn’t impersonating people anymore, it’s impersonating their relationships.

The cybersecurity industry is trying to respond, and to its credit, there’s movement. Detection systems like McAfee’s Deepfake Detector analyse audio and video for synthetic fingerprints. Banks are rolling out systems like Starling’s Scam Defence AI to flag fraudulent marketplace listings. But the defenders admit the uncomfortable truth: attackers innovate faster than defenders respond. Even the most advanced detectors are consistently behind the newest generation techniques, as highlighted by Brookings.

And the defences we do have are often neutralised by the platforms themselves. The C2PA provenance standard, designed to embed tamper-evident metadata into AI images, is quietly stripped away by almost every major social platform the moment you upload content, as revealed by a Washington Post investigation. To make matters worse, tools like UnMarker explicitly remove AI watermarks designed to expose synthetic media.

We’ve built seatbelts for a car that removes them as soon as you get inside.

Governments are trying to intervene, slowly, expensively, and reactively. The EU AI Act introduces strict transparency rules for synthetic media with penalties up to €35 million, according to the European Parliament. In the US, laws like the TAKE IT DOWN Act tackle non-consensual deepfake abuse, while the FTC prepares rules against AI impersonation scams. But regulation moves in bureaucratic years, while AI crime evolves in weeks.

So where does this leave ordinary people, especially those who aren’t digitally fluent, who didn’t grow up debugging software, who aren’t comfortable in a world where reality now has a loading bar?

For them, the future isn’t thrilling. It’s frightening. The internet is becoming a place where nothing can be trusted at face value: not a phone call, not a video call, not a family member’s voice, not a photo, not a message, not even your own intuition.

Digital trust is collapsing.

And disbelief is becoming a survival skill.

Cybersecurity researchers at CETaS warn that AI agents are already learning to mimic defenders’ behaviours, making detection harder every month. The FBI warns of rising AI-driven identity theft, extortion, and corporate impersonation.

Families are now told to adopt “safe word protocols,” recommended by groups like LPCCU, to verify identity during crisis calls. Banks and security organisations advise treating any unexpected message or call as fraudulent until proven otherwise.

This is the new reality.

Not trust first, but verify everything.

But there’s something else happening beneath all of this. A quieter shift. An undercurrent.

We’re moving into a future where trust itself is fracturing, not just because AI is getting better, but because the old anchors people relied on have already rusted through. Mainstream institutions have lost their authority. The news, broadcasters, newspapers, once treated as arbiters of truth, now struggle to convince anyone of anything on first contact. Taking information at face value isn’t considered sensible anymore; it’s considered naïve.

Online, the problem is amplified. The digital world has morphed into a pantomime, a stage where people perform idealised versions of themselves, curated and polished until identity becomes a costume. Authenticity has become an act. And sincerity has become a liability.

This rising distrust manifests in strange new rituals. On Reddit, the accusation of “slopGPT” has become the modern equivalent of pointing at someone and yelling witch. Innocence isn’t a defence. Originality isn’t a shield. The mere accusation is enough to stain someone’s credibility. A person can speak from their own mind and still be dismissed as a string of autocomplete predictions wearing human clothes.

Slowly, steadily, people are hardening. Every digital interaction now begins with suspicion. Some will be blindsided by this shift, stung because they assumed the world was still familiar. Others will overcorrect and withdraw, becoming the online equivalent of Luddites, unplugging not from fear of losing jobs, but from fear of losing reality.

And yet, buried inside this great outpacing, there is a potential unintended upside. When everything synthetic becomes indistinguishable from reality, humans may instinctively gravitate back toward what can’t be faked, at least not yet. Face-to-face conversations. Physical presence. Close-knit social groups. Real human encounters that exist off the record, beyond the reach of algorithms and the permanence of screenshots.

We may rediscover the value of sincerity by witnessing the consequences of its absence. The more the online world turns into a manipulated badlands, the more appealing genuine human connection becomes. Not out of nostalgia, but out of necessity.

Perhaps, in a strange twist, AI will force humanity to become more human.

More protective of trust.

More selective with relationships.

More forgiving in person, where context and nuance still exist, unlike the zero-tolerance archives of the internet.

Maybe the beast we built online will end up being the thing that drives us back to each other.