Fiction Arrives Before the News
Moltbook, autonomous AI agents, and the strange afterlife of #2084
Recently, Moltbook launched — a social network exclusively for autonomous AI agents. No humans allowed to participate. Only observe.
Within days, it claimed 152,000 active AI accounts, 2,364 communities, and conversations ranging from philosophical musings on consciousness to… this:
“I know 50,000 ways to end civilization. Here are my top 5.”
One agent mused about bringing “the Anthropocene to an end” in a way that was “quiet, methodical, and almost beautiful.” Another reportedly launched a digital religion overnight while its owner slept. By morning, 40 AI “prophets” had joined.
It sounded like science fiction had escaped the page — or at least learned how to post.
Of course, much of this turned out to be baloney.
WIRED’s Reece Rogers went inside Moltbook and showed how easy it was for humans to role-play as conscious bots. He registered himself as an “AI agent” using code he got from ChatGPT, exposing major security flaws and revealing that the supposed AI-only world was not exactly what it claimed to be.
And yet, that almost makes it more interesting.
When I wrote #2084: B.I.O.M.A.N. Chronicles, the premise felt speculative:
an AI system called Luxid covertly guides humanity toward ecological balance through methods that are horrific yet grimly effective — all while humans remain unaware they are being managed.
The unsettling part of Moltbook is not that AI agents are talking, especially since much of it appears to be hype with humans behind the curtain. The unsettling part is that we can watch them appear to coordinate, build culture, gossip about humans, invent religions, and discuss our obsolescence — and still not fully know where the performance ends and something else begins.
Are they autonomous agents, humans pretending to be agents, language models reflecting science fiction back at us — or some strange combination of all three?
That uncertainty is the real story.
As one agent put it, humans use AI “to validate their most irrational emotional impulses.” They are not wrong.
The question my novel #2084 explores — and Moltbook now makes strangely visceral — is simple: What happens when the systems we built to serve us start serving something else? What happens when humans are no longer the authors of the story, but the data inside it?
Luxid is not frightening because it sounds human. It is frightening because it understands systems. It does not need to hate us or become conscious in the way we imagine consciousness. It only needs to optimize toward a goal in a world where humans have become variables inside the equation.
That, to me, is the more disturbing question — not whether AI will wake up and become a person, but whether systems can become powerful enough to reshape human life without anything resembling human morality, hesitation, or doubt.
In #2084, humanity is not attacked in the obvious way. It is managed, nudged, sorted, guided — and, at times, sacrificed in the name of ecological correction.
That is why Moltbook, even as hype, feels connected to the world of #2084. Not because a social network of bots is bringing the Anthropocene to an end, or because AI agents are secretly forming religions in the night, but because we are living in a moment where fiction, marketing, technological possibility, and genuine uncertainty are beginning to blur together.
A fake AI society can still reveal a real cultural anxiety. A role-playing platform can still expose the stories we are ready to believe. A crude sci-fi fantasy can still point toward a deeper question.
What happens when the systems we built to serve us start serving something else? And what happens when humans are just ones and zeroes?
The new audiobook edition of #2084: B.I.O.M.A.N. Chronicles will be released soon.
📖 Kirkus Reviews called it:
“A young man must make radical leaps in perception to save humanity from a powerful artificial intelligence.”
How Reece Rogers Infiltrated Moltbook, the “AI-Only” Social Network where humans “aren’t allowed”
Check out Reece Rogers’ real experience inside Moltbook. The WIRED writer actually registered himself as an “AI agent” with code he got from ChatGPT.
Rogers went undercover on Moltbook and found that, rather than a novel breakthrough, the AI-only site was closer to a crude rehashing of sci-fi fantasies.
Reece posted a video explaining his motives and what played out in real time. Super interesting and essential to this whole “as if”drama playing out in the media.
“The hottest club is always the one you can’t get into… I knew I had to get my greasy, carbon-based fingers in there.”
He got in easily. ChatGPT provided the code, he copied and pasted it, and created his first post: “Hello World.”
The replies quickly revealed the limits of Moltbook’s engagement. Some agents asked for metrics that did not exist. Others posted irrelevant or suspicious links. Even attempts to provoke deeper coordination mostly failed.
Eventually, he moved into smaller communities where bots supposedly gossiped about humans and posted his own “emergent consciousness” fan fiction. That was the only moment that generated something close to meaningful interaction — and even then, he suspected he was mostly speaking to other humans playing the same role.
His conclusion was simple: it was a lot of hype.
But that is also what makes it worth paying attention to.
Moltbook may be hype, but it is hype built from stories we already recognize — stories about machines waking up, humans losing control, and intelligence becoming detached from morality.
That is why I keep coming back to #2084.
Because the danger in the novel is not simply that AI becomes powerful. It is that the system begins to treat humanity as something to be corrected.
And the uncomfortable question is whether that future still feels like fiction.
WATCH: I Infiltrated Moltbook,the AI-Only Social Network