As recently as nine months ago I felt I could tell the difference between AI and not-AI when it came to images and videos. That time has now passed. Today, every time I go online I have to wade through floods of AI slop before I find an island of meaning. My only guide is my knowledge of the wider context of the world. I make judgement calls (would Tom Hanks really sell a dental plan?) and decide with that context if what I’m looking at is real or fake. Sometimes the context is thin and the ad is convincing. It’s getting harder to navigate out there.
Deepfakes are images, videos or audio that have been edited or generated using AI to show someone doing or saying something they didn’t. Deepfakes used to be vanishingly rare, expensive to make and not that convincing. All of that has changed. AI is cheap and it is everywhere and our governments and legal systems haven’t begun to understand the repercussions, let alone to limit them.
In fairness, we’re all flailing. Things are changing fast.
Deepfakes can be most effective when they are personal and targeted. In 2024 a finance employee at an engineering company in Hong Kong entered what he thought was a video call. Multiple colleagues he recognized attended and took part in the meeting, including the company CFO, but everyone on the call was a deepfake. Following the meeting the finance employee made fifteen transfers totaling over twenty-five million dollars to accounts controlled by the scammers. No arrests were made and funds were never recovered.
And there are many other examples of more widely targeted frauds — paid ads on social media featuring deepfaked celebrities or high-profile business people spruiking ‘investments.’ Or deepfakes featuring politicians ‘caught’ doing something abhorrent and distributed far and wide by bot accounts.
This is rich territory for crime fiction.
In my book Three Reasons for Revenge, deepfakes are used as a means of personal attack. They appear twice in the story. The first is a video that appears to show a psychologist (Robert) having sex with a patient. The second is an audio recording that seems to expose a fashion designer (Elsa) as horribly judgmental and cruel about an influencer she has hired and publicly lauded. One of the recordings is fake, and one is real, but the psychologist and the fashion designer both immediately claim that the recording of them is a deepfake.
Which leaves the characters around them in an impossible situation. Do you believe what your eyes and ears are telling you? A video that appears to be utterly convincing? Or do you believe the person who, five minutes earlier, you would have trusted with your life?
A deepfake is the worst kind of character assassination if you are innocent. A video that uses the timbre and nuance and tone of your voice, that uses your lips and cheeks and laugh lines to convey a message that is the opposite of anything you would choose for yourself is a deep violence, not least because once that video is out in the world, you have to live in uncertainty. Always wondering who has watched it and who has believed it.
And the existence of deepfakes has created a credible defence for anyone now confronted with legitimate recordings. The ‘I’ve been deepfaked’ defence is 2026’s version of ‘I’ve been hacked.’ It can be used by anyone, but I expect that it will be used most powerfully by those with money. Because very wealthy people have the resources to hire PR firms who specialise in ‘reputation management’ which will in turn hire bot farms or influencers or clipping services to manipulate public conversation around a topic, and even turn the existence of a deepfake into an asset, rather than a liability.
The challenge for the characters in my book, and for all of us, is that when evidence itself becomes unreliable, accepted truth becomes less about what can be proven and more about who is believed, how often a claim is repeated and whose version of events reaches us first.
That was what interested me as a novelist. In Three Reasons for Revenge the deepfakes do more than threaten reputations. They fracture relationships. The people around Robert and Elsa must decide whether the trust built over years outweighs what they can see and hear in a few brutal seconds.
Crime novels have always asked how we know what happened, who we should believe, and what counts as evidence. And new technology has always brought opportunity and challenge to fictional and real-life policing. If the technology of deep-fakery is different, I think it is because it strikes at something that feels both personal and precious. Our trust in our own judgement, in our eyes and our ears. In each other.
***
