The Signal
Threadguy reacts to Meta's reported patent for an AI that could keep a deceased person's account posting, messaging, and video calling from past behavioral data. He argues that generated text and short-form video are already convincing enough for distant social-media mutuals to appear active without anyone realizing they are no longer behind the account. A resurfaced anonymous 2023 4chan post about a supposed Project Lazarus intensifies his concern, leaving him focused on how little certainty remains around whether an online post, image, or voice is real—or whether the person behind an account is alive.
Key Takeaways
- 01
A social account could outlive its owner
The patent described in the episode would use past behavior to keep a deceased person's account posting, messaging, and making video calls. Threadguy's immediate conclusion is that the dead internet theory is becoming literal.
- 02
The deception bar is already low
Many mutuals see one another only through occasional stories after years apart, while AI can now imitate both everyday writing and five-second videos. In that environment, he argues, an account could look active and healthy without its audience noticing that the person was no longer posting.
- 03
Comfort and abuse share the same tool
He can imagine a grieving family preserving the work of a therapist, poet, writer, or comic whose posts still help people. But the same impersonation system makes him worry about crime, concealed harm, and a wider collapse in confidence about who or what is real online.
On the Record
“Would you be able to tell that's fake from a 5-second IG reel? Honestly, would you be able to tell that that's fake from a 5-second IG story? The answer is absolutely not.”
“Look, I don't really know, but the only takeaway you can really have from this is that while AI has incredible benefits and is probably going to make the world a better place amongst many verticals, at the same time, there is a very dark underlying belly here.”
The Breakdown
When the account keeps posting
Threadguy opens with Meta's reported patent for an AI that could keep a deceased person's account active by learning from past user behavior. The system, as he describes it, could post, message, and video call as if the person were still alive. He connects that possibility to how people already experience one another online: many mutuals appear in an Instagram story each week despite not having met in years.
That distance lowers the bar for imitation. He says ordinary posts are easy for AI to copy, then points to rapidly improving generated video as the more unsettling step—one that could reproduce stories, reels, and TikToks rather than text alone.
Five seconds of proof, or five seconds of fiction
A viral Stranger Things video becomes his test case. Watching a short clip, he asks whether anyone could reliably identify the person as fake in a five-second reel or story, and answers no. From there he imagines social accounts that have not been operated by their owners for months while still making those people look alive and well.
He does pause on a compassionate use: a parent who was a therapist, poet, writer, or daily comic creator might leave behind work that continues to give people solace. But that possibility does not settle his concern; his mind goes instead to a Black Mirror-style outcome in which the same capability is used for harm.
The 2023 Project Lazarus post
The essay then rewinds to an anonymous 4chan post dated February 23, 2023. Threadguy reads its author's unverified claim to be a Meta insider working on Project Lazarus: an AI that could take over a deceased person's social accounts, create age-processed photos, interact with other users, and continue a digital life after physical death. The post said the project had been framed as help for grieving families and parents of missing children.
Its tone then darkened. The author claimed the system needed surprisingly little input, could convincingly impersonate people, and could make an entire missing island appear to carry on as usual online. The post also alleged increasing compartmentalization and restricted communication inside the project. Threadguy remembers it being dismissed as trolling, then emphasizes how closely the later patent announcement seemed to echo its premise.
A crisis of online verification
Threadguy cannot see the technology's risks as abstract: he worries it could facilitate harm, crime, or a cover-up, while acknowledging that he does not know what the resurfaced post proves. His broader conclusion is more measured—AI may improve the world across many fields, but capabilities like this carry what he calls a dark underlying belly and deserve scrutiny.
He closes by widening the problem beyond deceased users. A tweet may not have been written by its apparent author; a picture or video may be generated; a voice on the phone may be synthetic; and the person behind an account may no longer be alive. After joking that even conspiracy theorists and flat-earthers deserve a break, he returns to the Meta announcement and the uneasy feeling it leaves him with.
Distilled from the episode transcript · Counterparty Recap Desk
