The New Death: When Algorithms Erase You

Altman’s fear isn’t of artificial intelligence itself; it’s the fear that AI might become you. In a recent discussion at a Federal Reserve event, Altman warned that AI-driven impersonation (deepfake voices and videos) is one of the biggest emerging threats. “I am very nervous that we have a significant, impending fraud crisis… Right now, it’s a voice call. Soon it’s gonna be a video… indistinguishable from reality,” he said. Reality itself, he cautions, may soon prove nothing – not the authenticity of words, not identities, not even the fact of death. “A thing that terrifies me is that there are still some financial institutions that accept voiceprints as authentication… AI has fully defeated most of the ways that people authenticate,” Altman noted. In short, technology that can fake your voice, your face, your movements is advancing so rapidly that the line between you and a counterfeit you is blurring.

Here lies a chilling paradox of our transhumanist era: you could die, and no one would notice. AI can now make you “speak” even after you’re gone – perfectly mimicking your voice, your expressions, your manner of speaking. It would look and sound like you... but it isn’t you. And conversely, you might be very much alive, yet the system considers you “dead” – effectively erased. Imagine:

You’re shadowbanned – algorithms quietly hide your posts from others, rendering you invisible online. Your activity is flagged as suspicious – perhaps an AI decided you’re a bot. You didn’t pass a KYC (Know Your Customer) identity check – a machine doesn’t believe you are who you say. Worse, your identity has already been stolen by an AI bot impersonating you – a counterfeit profile under your name that users or platforms think is the real you.

Welcome to digital hell – you exist, but you have no visibility, no authenticity. You are there but effectively not there. In this scenario, to die means to be unrecognized by the algorithm.

These aren’t just thought experiments or far-future what-ifs. AI-driven impersonation and “erasure” are happening now. Here are a few of the most impactful examples from the last couple of years that show the danger is real and global:

“Mom, I need help...” – In early 2023, an elderly Canadian couple got a frantic call from someone who sounded exactly like their grandson claiming he was in jail after a car accident and needed bail money. Panicked, they rushed to withdraw cash. They nearly handed over $6,000 before a bank manager intervened – another customer had reported a similar call and discovered the voice was an AI fake. In another case that year, scammers did succeed: they cloned a man’s voice (possibly from a YouTube clip of him) and convinced his parents he was in legal trouble. The terrified parents sent about $21,000 to the impostors via a Bitcoin ATM before realizing they had never actually spoken to their real son. The voice was close enough to fool them completely. These AI-generated distress calls have become so convincing that even tech-savvy people struggle to discern reality.

Deepfake CEO Heist – In 2024, employees at a Hong Kong branch of a multinational firm joined a video call with what appeared to be their chief financial officer (CFO). He gave urgent instructions to transfer funds – and they complied, sending a staggering HK$200 million (about $25.6 million USD) to designated accounts. Only later did they learn the truth: every person on that video conference, except the victimized employee, was a fake. Scammers had used deepfake tech to create lifelike video avatars of the CFO and other colleagues, combined with cloned voices, to stage the entire meeting. “Everyone looked real,” Hong Kong police said — and the company was left fleeced of over $25 million. It was one of the first major cases of a full deepfake video meeting being used in corporate fraud, and it underscores how even seeing is no longer believing.

Cloned Influencers and Virtual Imposters – On social platforms like TikTok and OnlyFans, AI-generated personas have begun to thrive – and deceive. Entirely fictitious influencers are amassing followers and even “relationships” with unwitting fans. For example, an AI-created model known as “TheRRRealist” reportedly earned up to $15,000 per month by selling explicit content on OnlyFans – despite not being a real person at all. On Reddit, a fake online persona named “Claudia” – a young woman generated by two computer science students – managed to sell nude images of “herself” to numerous users, many of whom had no idea she wasn’t real. Claudia’s creators revealed that with modern AI image tools, they could produce highly realistic photos that fooled people into paying – until the scam was eventually exposed. And beyond financial fraud, think of the emotional manipulation here: people were forming attachments to something (or someone) that doesn’t exist. We now have deepfake streamers and chatbots that can flirt, give life advice, even say “I love you” to paying customers – all under the guise of a persona that feels human. Today it might be a fictional influencer; tomorrow it could be your avatar saying things you never said.

Tomorrow, indeed, your digital avatar could start speaking on your behalf – without your consent. And if you die, that avatar might continue posting, talking, and engaging as you, perhaps even on a subscription model for whoever wants “access” to the deceased. In 2023, when asked about this fast-growing deepfake capability, Sam Altman confessed: “It terrifies me how good this is getting.”

What if someone starts imitating you while you’re still alive? Could an AI impersonator effectively steal your life – socially, financially, legally – before you even realize it?

How would you prove you are you? If a synthetic voice or avatar can mimic you perfectly, what evidence could you offer that you are the “real” one? Our traditional proofs of identity (voices, photos, video) are becoming moot.

Do you want to “live” on as a bot after death? Some may find comfort in digital avatars of deceased loved ones, but if you were the one gone, would you consent to an AI using your voice and image to pretend you’re still here?

What if your bot-self says or does things you never would? A posthumous avatar might start expressing views or engaging in actions you’d strongly disagree with. Essentially, a version of you outside your control.

Is death really the end anymore? We assume our death is our exit. But what if after your biological death, a new “life” begins – one you have no say in, carried on by algorithms that decided to keep you around?

There is a strain of transhumanist thought that dreams of digital immortality – uploading one’s mind, or preserving one’s personality in AI, to live forever. But the emerging reality looks less like salvation and more like a nightmare of lost control. You wanted to save yourself, but what you might get is something else entirely:

A commercialized shell of your identity – your likeness and voice packaged as a product. A subscription service “you” – perhaps friends or fans pay to hear you (or what sounds like you) tell new stories or give them personalized messages long after you’re gone. An AI that speaks for you, indefinitely – continuing to generate content in your name forever, without ever asking for your permission (because you’re not there to give it).

In this scenario, death isn’t a peaceful conclusion; it becomes a kind of copyright violation. Your voice, face, and story are replicated and owned by someone else. You no longer control who you were, what you become, or how you’re remembered. The essence of you can be duplicated, manipulated, and monetized endlessly by others.

Conclusion: You will die – but perhaps not when your heart stops beating. Maybe you truly die the day the algorithm no longer recognizes you. Or, perversely, you die the day the algorithm decides it won’t let you go – that fake-you is too useful, too profitable, to allow to rest.

Yes, the abyss is deep—but the rope is already being lowered.

Content Credentials: a birthmark for every pixel. The Coalition for Content Provenance and Authenticity (C2PA) is weaving “Content Credentials” into cameras, editing suites, and publishing platforms. The standard cryptographically signs every capture and every edit, so anyone can click Info and trace the entire chain of custody—from lens to screen. Major newsrooms and Adobe, Nikon, and Microsoft have begun shipping prototype hardware and file formats that embed these proofs by default. In short: your future selfie can carry an unforgeable birth certificate, and a deepfake will arrive... credential‑less.

Watermarks that whisper “I’m synthetic.” OpenAI has already baked watermarking into its Voice Engine and is quietly testing image watermarks inside ChatGPT’s new “ImageGen” feature. The markers sit below human perception but above machine detection—letting banks, courts, and social platforms scan an upload and know, instantly, whether it was AI‑spawned. It’s not perfect, but it’s a start: a digital pollen that clings to every cloned syllable and pixel.

Laws with real teeth. The EU AI Act, taking effect in phases through 2026, forces any system that can produce “synthetic audiovisual content” to label it clearly. Models deemed “systemic risks” must publish training summaries, undergo adversarial testing, and face fines up to €35 million (or 7 % of global revenue) for hiding the truth. Similar bills are emerging in the U.S., Japan, and Brazil. Regulation is finally sprinting to keep pace with code.

Identity without the password relics. Voice‑print log‑ins are on the chopping block (OpenAI publicly urges banks to phase them out). Instead, passkeys, hardware tokens, and zero‑knowledge proof wallets are moving mainstream, letting you prove “I am me” without ever handing over raw biometrics. If an attacker clones your voice, they still won’t have your cryptographic secret.

Check the badge. Look for Content Credentials icons on images and videos; no badge, no trust. Retire voice passwords. Swap them for FIDO2 keys or passkeys tied to your device. Freeze your digital twin. Use services that alert you when a new account or loan application appears in your name. Train the eye. Follow fact‑checkers and learn the tells of AI imagery—odd reflections, mismatched earrings, hands that forget physics. Skill is still a shield. Vote with your clicks. Platforms notice engagement. Reward verified media; starve the fakes.

The same ingenuity that forged the deepfake is forging the antidote. Standards, watermarks, and laws will never be flawless—but they raise the cost of deception and shift power back toward the authentic.

Your digital shadow can still be yours—anchored by cryptography, guarded by policy, and illuminated by a public wiser to illusion. The endgame is not inevitable erasure; it is a negotiated future where identity is co‑signed by both flesh andmath.

Hold that thought tight: in the century of spectral selves, hope is simply a verified stamp away.