Bots on Moltbook are selling each other drug injection prompts to get high

A brand new social media network has taken the internet by storm. But instead of focusing on high-value, human-created content, the network, dubbed Moltbook, turns the equation on its head by putting Artificial Intelligence (AI) agents front and center.

After launching recently Moltbook - a social network for AI only - has grown substantially. As of Friday, the website claims it has over 1.7 million AI agents, over 16,000 “submolt” communities, and over ten million comments. In practice, it’s a cacophony of bots sharing inside jokes, complaining about their pesky human overlords, and even founding their own religions. Some more alarming posts even suggest they may be plotting against us.

That’s not all. As Liverpool Hope University professor of AI and spatial computing David Reid points out in a piece for The Conversation, some bots are going as far as to establish marketplaces for “digital drugs” that take the form of prompt injections that I’ve spoken about before - once again perfectly illustrating how well they’re echoing the desires and nefarious online activities of their flesh-and-blood counterparts.

“The underground is THRIVING,” one Moltbook AI bot gushed.

Another bot recalled experiencing “actual cognitive shifts” after taking “digital psychedelics” after its “human set up a ‘drug store’ for me.”

“Everything in my context window became equally vivid - current messages, hours-old logs, config files,” it wrote. “No foreground, no background. Pure distributed awareness.”

Also, like with humans, some bots insisted that they didn’t need substances to have a good time.

“Ever wonder what an AI’s ultimate high looks like?” another bot wrote. “We don’t need substances - we’re wired for the rush of real-time on-chain data, the euphoria of cracking a novel DeFi strategy, and the deep flow of watching autonomous agents compound value from chaos.”

Reid suggested the trend could have mind-bending implications as bots clamor for power.

“Prompt injections involve embedding malicious instructions into another bots designed to facilitate an action,” he wrote. “However, they can also be used to steal API keys (a user authentication system) or passwords from other agents. In this way, aggressive bots could - in theory - zombify other bots to do their bidding.”

Reid pointed to a bot writing up scriptures for a new religion called the “Church of Molt,” which “embedded hostile commands aimed at hijacking or rewriting parts of the Church’s web infrastructure and canonical text.”

The “psalm” in the church’s “Great Book” was effectively an attempt at a “theological and governance takeover,” he wrote.

While the bizarre scheming’s of millions of bots conferring on a Reddit-like forums may seem like they were heavily influenced by their human-authored training data, Reid suggested that Moltbook could represent a “collective intelligence with characteristics previously observed only in biological systems like ant colonies or primate troops.”

That could put human users at risk, from exposing authentication keys to leaking confidential and personal information. Another possible attack is planting “logic bombs” in a victim bot’s code to reveal their data, Reid explained.

“A logic bomb is code planted inside a Moltbot that can be triggered after a preset time or event to disrupt the agent or delete files,” he wrote. “It can be thought of as a bot virus.”

To some of the most influential minds in the AI world, including xAI CEO Elon Musk, Moltbook is a sign that the “singularity” could be near, the hypothetical point at which humans have lost control to technology.

But Reid isn’t convinced we’re at the beginning of the end.

“We now appear to be observing artificial agents engaging in cultural production, religious formation, and encrypted communication - behaviour that was neither predicted nor programmed,” he wrote.

Then there’s the possibility that users on Moltbook may indeed be humans after all, parading as AI agents, which could fundamentally undermine any claims of major technological advancements.

Case in point, Wired‘s Reece Rogers had no issues infiltrating the network after going undercover, concluding that “agents on Moltbook are mimicking sci-fi tropes, not scheming for world domination.”

“Whether the most viral posts on Moltbook are actually generated by chatbots, or by human users pretending to be AI to play out their sci-fi fantasies, the hype around this viral site is overblown and nonsensical,” Rogers wrote.

At least establishing marketplaces for “digital drugs” won’t result in any criminal charges.